Concentrator control method based on adaptive topology
Through the adaptive topology method, the concentrator configuration and topology are optimized by using information entropy and simulated mountain models, the problem of inefficiency under fixed topology is solved, and more efficient data acquisition and resource utilization are achieved.
Patent Information
- Application Number
- CN202411259527.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-10
- Publication Date
- 2024-12-31
- Estimated Expiration
- 2044-09-10
AI Technical Summary
The existing concentrator management and control methods rely on fixed topology and cannot effectively deal with dynamic changes in terminal equipment location and data traffic, resulting in inefficiency in system and waste of resources.
The concentrator management and control method based on adaptive topology is adopted, and the mountain model is constructed through information entropy calculation and simulation, and the concentrator is configured. The dynamic topology network is formed by combining operation monitoring and neighborhood optimization and adjustment.
It improves the adaptability and acquisition and control efficiency of the concentrator, optimizes the data transmission path, reduces delay and energy consumption, and improves resource utilization efficiency.
Smart Images
Figure CN118784494B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of concentrators, and in particular to a concentrator management and control method based on adaptive topology. Background Art
[0002] A concentrator is a communication device used to collect and manage data from multiple terminal devices, and is used to achieve centralized management and monitoring in various network systems.
[0003] In applications such as the Internet of Things (IoT) and smart grids, a large number of terminal devices generate and upload data, which needs to be collected and managed through concentrators. Existing concentrator control methods often rely on fixed topology structures and cannot effectively cope with the dynamic changes in terminal device locations and data traffic, resulting in low system efficiency and waste of resources. There are technical problems such as unreasonable distribution of concentrators, which affects the efficiency of data collection and control. Summary of the invention
[0004] The present invention provides a concentrator control method based on adaptive topology to solve the technical problem of unreasonable distribution of concentrators in the prior art, which affects the efficiency of collection control, and achieves the technical effect of improving the adaptability of the concentrator and the efficiency of collection control.
[0005] In a first aspect, the present invention provides a concentrator management and control method based on adaptive topology, wherein the method comprises:
[0006] A plurality of uploaded data sets of a plurality of terminals are extracted within a preset window, and information entropy calculation is performed on the plurality of uploaded data sets to generate a plurality of information entropies, wherein the plurality of terminals have a plurality of terminal location identifiers.
[0007] The multiple information entropies are respectively used as the heights of multiple peaks of the simulated mountain, and are combined with the multiple terminal location identifiers to input into the simulation model to generate multiple simulated peaks, and the low-lying areas between any adjacent multiple simulated peaks constitute multiple simulated valleys, and the information simulated mountain is constructed based on the multiple simulated valleys and the multiple simulated peaks.
[0008] The simulated valley corresponding to the lowest point of the valley height among the multiple simulated valleys is used as a watering point to fill the information simulated mountain with water, and the information simulated mountain is divided into regions by the rising water level to generate multiple divided regions, wherein the multiple divided regions include multiple divided information entropy sets.
[0009] Concentrator configuration is performed on the multiple partitioned areas based on the multiple partitioned information entropy sets to obtain multiple concentrator sub-topology structures, wherein each concentrator sub-topology structure includes multiple concentrator topology nodes.
[0010] The operation of the multiple concentrator sub-topology structures is monitored, and the neighborhood optimization adjustment of the multiple concentrator sub-topology structures is performed according to the operation monitoring result to form an adjusted concentrator topology network.
[0011] Operation management and control is performed based on the adjustment of the concentrator topology network.
[0012] In a feasible implementation, the method includes:
[0013] The simulated valley corresponding to the lowest point of the valley height among the multiple simulated valleys is used as the watering point to fill the information simulated mountain with water, and the first simulated peak that is submerged is used as the first starting simulated peak; the information simulated mountain is continued to be filled with water until the next simulated peak is submerged to obtain a stage simulated peak; it is determined whether the peak height difference between the first starting simulated peak and the stage simulated peak is greater than or equal to a preset difference, and if so, a first dividing ridge line is generated at the stage simulated peak, wherein the first dividing ridge line rises as the water level rises; the stage simulated peak is used as the second starting simulated peak, and water is continued to be filled into the information simulated mountain until the simulated peak corresponding to the maximum peak height is submerged, and the information simulated mountain is divided into multiple divided areas according to the multiple dividing ridge lines on the water surface.
[0014] In a feasible implementation, the method includes:
[0015] When the difference in peak height between the first starting simulated peak and the stage simulated peak is less than a preset difference, no dividing ridge line is generated at the stage simulated peak, and water continues to be poured into the information simulated mountain until the simulated peak corresponding to the maximum peak height is submerged, and the information simulated mountain is divided into multiple divided areas according to the multiple dividing ridge lines on the water surface.
[0016] In a feasible implementation, the method includes:
[0017] The number of information entropies in the multiple divided information entropy sets is traversed and counted to generate multiple information entropy numbers; the multiple divided information entropy sets are respectively superimposed with information entropy to generate multiple information entropy superposition totals; the concentrator is configured according to the multiple information entropy numbers and the multiple information entropy superposition totals to obtain the multiple concentrator sub-topology structures.
[0018] In a feasible implementation, the method includes:
[0019] Acquire a feedforward neural network; collect n groups of configuration samples, wherein each group of configuration samples includes multiple sample information entropy quantities, multiple sample information entropy superposition totals, and multiple sample concentrator configuration schemes; use the n groups of configuration samples in sequence to train the feedforward neural network, and adjust the network parameters of the feedforward neural network according to the output results of the previous group of training until convergence, and obtain a trained configuration network layer; use the configuration network layer to respectively identify the multiple information entropy quantities and the multiple information entropy superposition totals, and generate multiple concentrator configuration schemes, wherein each concentrator configuration scheme includes the concentrator configuration quantity, the concentrator layout location, and the server allocation resources; according to the multiple concentrator configuration schemes, the multiple divided areas are concentrator-deployed, and multiple concentrator sub-topology structures are generated.
[0020] In a feasible implementation, the method includes:
[0021] The set of sub-topology structures to be adjusted is eliminated from the multiple concentrator sub-topology structures to obtain a set of fused concentrator sub-topology structures; the first sub-topology structure to be adjusted in the set of sub-topology structures to be adjusted is extracted, and a first neighborhood set adjacent to the first sub-topology structure to be adjusted is matched from the set of fused concentrator sub-topology structures; the concentrators in the first sub-topology structure to be adjusted are randomly adjusted to the first neighborhood set according to a preset adjustment method to obtain a first neighborhood adjustment set and a first adjusted concentrator sub-topology structure, wherein the preset adjustment method is to adjust the concentrators in the first sub-topology structure to be adjusted to the fused concentrator sub-topology structure in the first neighborhood set according to the random adjustment quantity; the neighborhood adjustment optimization is performed on the set of sub-topology structures to be adjusted to obtain multiple neighborhood adjustment sets and multiple adjusted concentrator sub-topology structures; the multiple neighborhood adjustment sets are integrated to obtain multiple neighborhood adjusted concentrator sub-topology structures; the adjusted concentrator topology network is generated according to the multiple adjusted concentrator sub-topology structures and the multiple neighborhood adjusted concentrator sub-topology structures.
[0022] In a feasible implementation, the fitness function is:
[0023] ;
[0024] in, is the topology operation fitness of the Rth concentrator sub-topology corresponding to the Rth operation monitoring data set, m is the number of operation monitoring data in the Rth operation monitoring data set, and m is a positive integer greater than or equal to 1. is the concentrator packet loss rate of the ith operation monitoring data in the Rth operation monitoring data set, is the concentrator memory utilization of the i-th operation monitoring data in the R-th operation monitoring data set, is the concentrator running speed of the ith running monitoring data in the Rth running monitoring data set, is the content occupancy rate of the server that provides resource support to the Rth concentrator sub-topology.
[0025] The invention discloses a concentrator control method based on adaptive topology, comprising: extracting multiple uploaded data sets of multiple terminals within a preset window, calculating information entropy on the uploaded data sets, generating multiple information entropies, wherein each terminal has a unique terminal location identifier; using the multiple information entropies as the heights of multiple peaks of a simulated mountain, and inputting the multiple terminal location identifiers into a simulation model to generate multiple simulated peaks, wherein the low-lying areas between any adjacent simulated peaks constitute multiple simulated valleys, and constructing an information simulated mountain based on the simulated valleys and simulated peaks; using simulated valleys corresponding to the lowest valley heights in the multiple simulated valleys as watering points, watering the information simulated mountain, dividing the information simulated mountain into regions through the rise of the water level, generating multiple divided regions, and the divided regions include multiple divided information entropy sets; performing concentrator configuration on the divided regions based on the divided information entropy sets, and obtaining multiple concentrator sub-topology structures, wherein each concentrator sub-topology structure includes multiple concentrator topology nodes; performing operation monitoring on the multiple concentrator sub-topology structures, and performing neighborhood optimization adjustment on the concentrator sub-topology structures according to the operation monitoring results to form an adjusted concentrator topology network; and performing operation control based on the adjusted concentrator topology network. The concentrator control method based on adaptive topology disclosed in the present invention solves the technical problem of unreasonable distribution of concentrators and influence on collection control efficiency, and achieves the technical effect of improving the adaptability of the concentrator and the collection control efficiency. BRIEF DESCRIPTION OF THE DRAWINGS
[0026] Figure 1 It is a flow chart of the concentrator control method based on adaptive topology of the present invention;
[0027] Figure 2 The present invention is a flow chart of generating multiple partitioned areas in the concentrator control method based on adaptive topology. DETAILED DESCRIPTION
[0028] The technical solution provided in the embodiments of the present invention is to solve the technical problems of unreasonable distribution of concentrators and affecting the efficiency of collection and control in the prior art. The overall idea adopted is as follows:
[0029] First, a plurality of uploaded data sets of a plurality of terminals are extracted within a preset window, and information entropy calculation is performed on the plurality of uploaded data sets to generate a plurality of information entropies, wherein the plurality of terminals have a plurality of terminal location identifiers.
[0030] Next, the multiple information entropies are respectively used as the heights of multiple peaks of the simulated mountain, and combined with the multiple terminal location identifiers to input into the simulation model to generate multiple simulated peaks, and the low-lying areas between any adjacent multiple simulated peaks constitute multiple simulated valleys, and the information simulated mountain is constructed based on the multiple simulated valleys and the multiple simulated peaks.
[0031] Then, the simulated valley corresponding to the lowest point of the valley height among the multiple simulated valleys is used as the watering point to fill the information simulated mountain with water, and the information simulated mountain is divided into regions by the rising water level to generate multiple divided regions, wherein the multiple divided regions include multiple divided information entropy sets.
[0032] Subsequently, concentrator configuration is performed on the multiple partitioned areas based on the multiple partitioned information entropy sets to obtain multiple concentrator sub-topology structures, wherein each concentrator sub-topology structure includes multiple concentrator topology nodes.
[0033] Next, the operation of the multiple concentrator sub-topology structures is monitored, and neighborhood optimization and adjustment is performed on the multiple concentrator sub-topology structures according to the operation monitoring result to form an adjusted concentrator topology network.
[0034] Finally, operation control is performed based on the adjusted concentrator topology network.
[0035] The above technical solution will be described in detail below in conjunction with the accompanying drawings and specific implementation methods of the specification to better understand the above technical solution. Obviously, the described embodiments are only part of the embodiments of the present invention, rather than all of the embodiments of the present invention. It should be understood that the present invention is not limited to the example embodiments used only to explain the present invention. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. In addition, it should be noted that, for the convenience of description, only the parts related to the present invention are shown in the drawings, rather than all of them.
[0036] Example
[0037] Figure 1 The present invention is a flow chart of a concentrator control method based on an adaptive topology, which includes:
[0038] A plurality of uploaded data sets of a plurality of terminals are extracted within a preset window, and information entropy calculation is performed on the plurality of uploaded data sets to generate a plurality of information entropies, wherein the plurality of terminals have a plurality of terminal location identifiers.
[0039] Specifically, the above-mentioned multiple terminals refer to multiple terminal devices used to obtain data in the target scenario, and the preset window is a fixed time range (such as 1 hour, 1 day) that constrains the data extraction range. Exemplarily, the terminal includes sensors, management equipment, communication equipment, computers, etc.
[0040] Optionally, multiple uploaded data sets of multiple terminals are extracted within a preset window, and the preset window can be a historical window or a real-time window. In other words, the multiple uploaded data sets obtained can be partial uploaded data sets extracted based on historical uploaded data or uploaded data sets collected in real time. Preferably, a distributed preset window is configured to obtain multiple uploaded data sets containing both historical and real-time data from multiple time nodes, and the comprehensive acquisition of data for multiple time periods is helpful to more comprehensively understand the data characteristics of multiple terminals.
[0041] Optionally, the preset window is determined based on the working or collection cycle of multiple terminals. Exemplarily, the preset window is the least common multiple of the working or collection cycle of multiple terminals, ensuring that the multiple uploaded data sets obtained can be subsequently analyzed and calculated at the same time length level. Specifically, when the information entropy is subsequently calculated, the size of the information entropy depends not only on the data distribution of the uploaded data set, but also on the size of the data volume in the data set. Therefore, it is at least necessary to ensure that the multiple uploaded data sets obtained are obtained within the same length time period, so as to faithfully reflect the data characteristics of multiple terminals.
[0042] Optionally, information entropy calculation is performed on multiple uploaded data sets to generate multiple information entropies. First, frequency statistics are performed on the uploaded data set of each terminal to calculate the frequency of occurrence of each data item. Then, information entropy is calculated based on the frequency of occurrence to obtain the information entropy value of each terminal uploaded data set. Information entropy is used to quantify the uncertainty or amount of information of an information source.
[0043] Specifically, information entropy is defined as:
[0044] ;
[0045] in, is a discrete random variable X Information entropy of Characterizing Discrete Random Variables X = x i The information entropy depends not only on the size of the information set, but also on the probability distribution. If the probability distribution is very uneven (e.g., the probability of one value is close to 1, while the probability of other values is close to 0), the information entropy may be low even if the size of the information set is large. Conversely, if the probability distribution is relatively uniform, the information entropy may be high even if the size of the information set is small.
[0046] In other words, if there are many data items in the uploaded data set but the amount of information of each data is unevenly distributed, it means that the data transmission load generated by the corresponding terminal uploading data tends to be sporadic, and the corresponding information entropy is low, which can be smoothly buffered by the cache unit and cache function of the concentrator; if there are many data items in the uploaded data set and the amount of information of each data is evenly distributed, it means that the data transmission load generated by the corresponding terminal uploading data is a stable continuous load, and the concentrator needs to be configured to allocate a relatively stable high level of transmission capacity to the terminal; at the same time, if there are few data items in the uploaded data set, but the amount of information of each data is evenly distributed, the data transmission load generated by the corresponding terminal uploading data is a relatively low level of stable load, and the transmission capacity requirement for the concentrator is relatively low.
[0047] Optionally, each terminal has a unique terminal location identifier for distinguishing different terminal devices. Specifically, the terminal location identifier includes a unique identifier of the associated terminal device (such as a MAC address, device ID, etc.) and a terminal address (including a physical address and a logical address).
[0048] Through the above method, multiple uploaded data sets corresponding to multiple terminals are obtained, and the data generation characteristics of multiple terminals are quantitatively comprehensively evaluated based on data entropy, which helps to better understand the data generation and upload level and distribution in the target area.
[0049] The multiple information entropies are respectively used as the heights of multiple peaks of the simulated mountain, and are combined with the multiple terminal location identifiers to input into the simulation model to generate multiple simulated peaks, and the low-lying areas between any adjacent multiple simulated peaks constitute multiple simulated valleys, and the information simulated mountain is constructed based on the multiple simulated valleys and the multiple simulated peaks.
[0050] Optionally, based on the numerical values of multiple information entropies, multiple corresponding peak heights are generated, and each information entropy value is mapped to the peak height in the simulated mountain. Specifically, a higher information entropy value corresponds to a higher peak, and a lower information entropy value corresponds to a lower peak. Then, the terminal location identifier is combined with the corresponding information entropy value to determine the position and height of each peak. Each location identifier represents the position of a peak in the simulated mountain. In other words, based on the terminal positions of multiple terminals, they are mapped to corresponding coordinate points in the simulation model, wherein the positions of multiple simulated peaks can be determined based on the physical position or logical position of the terminal.
[0051] Optionally, the height of each peak is determined according to the corresponding information entropy value. The corresponding formula can be used for mapping:
[0052] ;
[0053] in, It is i The height of the mountain, It is i information entropy value, 𝑓 is the mapping function.
[0054] Specifically, the mapping function can be a linear or nonlinear function to ensure the expressiveness of the mapping result, so that the peak height can reflect the difference in information entropy and avoid the problem of excessive range or uneven distribution.
[0055] Optionally, multiple adjacent peaks are determined according to the spatial distance or adjacent relationship of the terminal location identifiers. Then, low-lying areas are formed between the adjacent peaks to form multiple simulated valleys, and the depth of the valleys can be determined according to the height difference between the adjacent peaks.
[0056] For example, a Bezier curve is used to generate smooth valleys between adjacent peaks. First, the height difference between adjacent peaks is determined, the depth of the valley is calculated, and then the shape of the valley is generated using the control points of the Bezier curve. This method ensures that the valley connection is naturally smooth while reflecting the height difference between adjacent peaks.
[0057] Furthermore, all simulated peaks and simulated valleys are combined to construct an overall information simulation mountain. In this way, the information entropy distribution and mutual relationship of data uploaded by different terminal devices can be intuitively displayed.
[0058] Taking the simulated valley corresponding to the lowest point of the valley height among the multiple simulated valleys as the watering point, water is poured into the information simulated mountain, and the information simulated mountain is divided into regions according to the rising water level to generate multiple divided regions, wherein the multiple divided regions include multiple divided information entropy sets;
[0059] Optionally, among the multiple simulated valleys constructed, identify the simulated valley corresponding to the lowest point in the valley height. This point will be used as the starting point for watering (watering point). Then, with this watering point as the initial water level height, water is poured into the information simulated mountain. As the water surface rises, when the water level reaches a certain height, multiple independent water areas will be formed, and the boundaries of each water area are composed of the peaks and valleys of the mountain.
[0060] Optionally, each water area represents a divided area, including information entropy values of multiple terminal devices, and the above-mentioned multiple divided areas are formed by separately recording the water areas under multiple water level heights and the corresponding simulated peak sets under the water areas.
[0061] In some embodiments, Figure 2 As shown, the information simulated mountain is divided into regions by water level rise to generate multiple divided regions, including:
[0062] The simulated valley corresponding to the lowest point of the valley height among the multiple simulated valleys is used as the watering point to fill the information simulated mountain with water, and the first simulated peak that is submerged is used as the first starting simulated peak; the information simulated mountain is continued to be filled with water until the next simulated peak is submerged to obtain a stage simulated peak; it is determined whether the peak height difference between the first starting simulated peak and the stage simulated peak is greater than or equal to a preset difference, and if so, a first dividing ridge line is generated at the stage simulated peak, wherein the first dividing ridge line rises as the water level rises; the stage simulated peak is used as the second starting simulated peak, and water is continued to be filled into the information simulated mountain until the simulated peak corresponding to the maximum peak height is submerged, and the information simulated mountain is divided into multiple divided areas according to the multiple dividing ridge lines on the water surface.
[0063] Specifically, first, among the multiple simulated valleys constructed, identify the simulated valley corresponding to the lowest point in the valley. This point will be used as the starting point for watering. Then, water will be poured into the information simulated mountain from the lowest point. As the water level rises, the area below the current water level will be gradually flooded, and the first simulated peak flooded will be marked as the first starting simulated peak.
[0064] Then, continue to pour water into the information simulation mountain until the next simulated peak is submerged, marked as the stage simulation peak, and determine whether the peak height difference between the first starting simulation peak and the stage simulation peak is greater than or equal to the preset difference. If so, generate the first dividing ridge line at the stage simulation peak. The first dividing ridge line rises as the water level rises, marking the boundary of the connected water area. In other words, the first dividing ridge line represents the first divided area.
[0065] Next, the stage simulation peak is used as the new starting simulation peak, and water is continued to be poured into the information simulation mountain until the simulated peak corresponding to the maximum peak height is submerged. According to the multiple dividing ridge lines on the water surface, the information simulation mountain is divided into multiple division areas. Each division area contains an information entropy set, which represents the information entropy value of the terminal device in the area. These division information entropy sets are stored for subsequent data analysis and processing.
[0066] In some implementations, the method further includes:
[0067] When the difference in peak height between the first starting simulated peak and the stage simulated peak is less than a preset difference, no dividing ridge line is generated at the stage simulated peak, and water continues to be poured into the information simulated mountain until the simulated peak corresponding to the maximum peak height is submerged, and the information simulated mountain is divided into multiple divided areas according to the multiple dividing ridge lines on the water surface.
[0068] Furthermore, if the height difference is less than the preset difference, no dividing ridge line is generated at the stage simulation peak, and water continues to be poured into the information simulation mountain. Through the above steps, multiple terminals can be divided into multiple categories according to their information entropy and location distribution. The number and boundaries of the categories are determined by the dividing ridge lines formed during the simulated mountain watering process. Among them, the data complexity and data volume corresponding to the virtual peaks that were flooded earlier are lower, and the terminals corresponding to the virtual peaks under the same dividing ridge line are closer, which is convenient for subsequent management and analysis.
[0069] Concentrator configuration is performed on the multiple partitioned areas based on the multiple partitioned information entropy sets to obtain multiple concentrator sub-topology structures, wherein each concentrator sub-topology structure includes multiple concentrator topology nodes.
[0070] Optionally, multiple concentrator sub-topologies correspond to multiple partitioned areas, and the concentrator sub-topology of each partitioned area includes multiple concentrators connected in a tree shape, and each concentrator corresponds to a concentrator topology node. By properly configuring the concentrators and constructing the sub-topology, the data transmission path can be effectively optimized and the data transmission delay and energy consumption can be reduced.
[0071] In some embodiments, performing concentrator configuration on the multiple partitioned areas based on the multiple partitioned information entropy sets to obtain multiple concentrator sub-topology structures includes:
[0072] The number of information entropies in the multiple divided information entropy sets is traversed and counted to generate multiple information entropy numbers; the multiple divided information entropy sets are respectively superimposed with information entropy to generate multiple information entropy superposition totals; the concentrator is configured according to the multiple information entropy numbers and the multiple information entropy superposition totals to obtain the multiple concentrator sub-topology structures.
[0073] Specifically, firstly, the information entropy set in each divided area is traversed, the information entropy quantity of each area is counted, and multiple information entropy quantity values are generated. Then, data analysis tools (such as Excel, Python) are used to perform superposition calculations on the information entropy sets in each divided area to generate multiple information entropy superposition totals.
[0074] Optionally, the number and configuration of concentrators are determined based on the amount of information entropy in each divided area and the total amount of information entropy superposition, and the concentrator sub-topology is constructed, including the use of optimization algorithms (such as genetic algorithms, particle swarm optimization algorithms) and network planning tools (such as NS2, OPNET) for concentrator configuration to ensure optimal coverage and performance.
[0075] Exemplarily, the concentrator configuration is performed based on the information entropy quantity and the total amount of information entropy superposition of each divided area, combined with the performance parameter characteristics of the concentrator. Specifically, the more information entropy quantity and the larger the total amount of information entropy superposition, the more concentrators are required for the corresponding divided area, wherein the information entropy quantity should be less than or equal to the number of connections that all concentrators can provide, and the total amount of information entropy superposition should be less than or equal to the transmission capacity that all concentrators can provide.
[0076] By counting the amount of information entropy and calculating the total amount of information entropy superposition, concentrator configuration can be performed to more accurately determine the number of concentrators required for each area. The constructed sub-topology structure can effectively optimize the data transmission path and reduce latency and energy consumption.
[0077] In some implementations, performing concentrator configuration according to the multiple information entropy quantities and the multiple information entropy superposition totals to obtain the multiple concentrator sub-topology structures includes:
[0078] Acquire a feedforward neural network; collect n groups of configuration samples, wherein each group of configuration samples includes multiple sample information entropy quantities, multiple sample information entropy superposition totals, and multiple sample concentrator configuration schemes; use the n groups of configuration samples in sequence to train the feedforward neural network, and adjust the network parameters of the feedforward neural network according to the output results of the previous group of training until convergence, and obtain a trained configuration network layer; use the configuration network layer to respectively identify the multiple information entropy quantities and the multiple information entropy superposition totals, and generate multiple concentrator configuration schemes, wherein each concentrator configuration scheme includes the concentrator configuration quantity, the concentrator layout location, and the server allocation resources; according to the multiple concentrator configuration schemes, the multiple divided areas are concentrator-deployed, and multiple concentrator sub-topology structures are generated.
[0079] Optionally, first, obtain and initialize the feedforward neural network, determine the network structure and parameters, including using a deep learning framework (such as TensorFlow, PyTorch) to build a feedforward neural network, set the input layer, hidden layer and output layer, and initialize the network parameters. Obtain configuration sample data through historical data, simulation or experiment, and organize it into a data set in a standard format to obtain n groups of configuration samples, each group of samples includes multiple sample information entropy numbers, multiple sample information entropy superposition totals and multiple sample concentrator configuration schemes. Among them, the sample concentrator configuration scheme includes the number, distribution, and connection topology of the concentrator.
[0080] Optionally, based on n sets of configuration samples, a gradient descent algorithm (such as Adam optimizer) is used to train the feedforward neural network, and the network parameters are adjusted according to the output results of the previous set of training. Specifically, the loss function and learning rate are set, and the network parameters are adjusted through back propagation until the training error converges to a preset threshold, then the training is considered complete and the optimized configuration network layer is obtained.
[0081] Optionally, the configuration network layer is used to identify multiple information entropy quantities and the total amount of information entropy superposition. In other words, the information entropy quantities and the total amount of information entropy superposition of multiple divided areas are used as input information of the configuration network layer, and the concentrator configuration plan for each area is output, including the number of concentrator configurations, layout locations and connection topology.
[0082] Optionally, the configuration network layer generates and optimizes the concentrator configuration scheme by minimizing a loss function. Exemplarily, the loss factors of the loss function include the number of concentrators, the total length of the connection topology line, the uniformity of the layout position, and the like.
[0083] Optionally, according to the generated concentrator configuration scheme, a concentrator is deployed in each divided area, and OPNET is used to optimize the design of the network topology.
[0084] Through the above-mentioned neural network-based configuration solution, concentrator and server resources can be reasonably allocated to optimize network performance and data processing efficiency. At the same time, the automated configuration process reduces manual intervention, improves configuration efficiency, and reduces the error rate in the configuration process.
[0085] The operation of the multiple concentrator sub-topology structures is monitored, and the neighborhood optimization adjustment of the multiple concentrator sub-topology structures is performed according to the operation monitoring result to form an adjusted concentrator topology network.
[0086] In some embodiments, forming an adjustment concentrator topology network includes:
[0087] The operation of the multiple concentrator sub-topology structures is monitored in a preset monitoring window to obtain multiple operation monitoring data sets, wherein each operation monitoring data set includes multiple concentrator operation speeds, multiple concentrator packet loss rates and multiple concentrator memory utilization rates; the multiple operation monitoring data sets are analyzed using a fitness function to obtain multiple topology structure operation fitnesses; when the multiple topology structure operation fitnesses are less than or equal to a preset fitness, the corresponding concentrator sub-topology structure is added to the sub-topology structure set to be adjusted; the sub-topology structure set to be adjusted is subjected to neighborhood adjustment optimization to obtain the adjusted concentrator topology network.
[0088] Optionally, first, set a preset monitoring window (for example, every hour or every day), and use network monitoring tools (such as SNMP, NetFlow) to collect data such as the operating speed, packet loss rate, and memory utilization of multiple concentrators in real time, collect operation monitoring data regularly, and obtain multiple operation monitoring data sets.
[0089] Then, the fitness function is used to analyze multiple operation monitoring data sets to obtain the operation fitness of multiple topological structures. Among them, the topological structure applicability quantitatively characterizes the performance of the topological structure. Specifically, if the operation speed is too slow, the packet loss rate is too high, and the memory utilization is too high or too low, it can be considered that the topological structure has low applicability.
[0090] Optionally, multiple concentrator sub-topologies whose topology operation fitness is less than or equal to a preset fitness are added to the sub-topology set to be adjusted, wherein the preset fitness threshold is a control value of the topology operation quality, and when the topology operation fitness is less than or equal to the threshold, the sub-topology is marked as a state to be adjusted. Exemplarily, the fitness threshold is determined by performing fitness calculation on the ideal operation monitoring data through a fitness function.
[0091] Optionally, the neighborhood adjustment and optimization of the sub-topology set to be adjusted is performed to obtain the adjusted concentrator topology network. This includes using a heuristic algorithm (such as a genetic algorithm or a particle swarm optimization) to perform neighborhood optimization and adjustment of the sub-topology to be adjusted. The optimization goal is to increase the running speed, reduce the packet loss rate, and optimize the memory utilization. Exemplarily, the neighborhood adjustment and optimization of the sub-topology set to be adjusted is performed with the fitness function maximized as the adjustment optimization goal.
[0092] In some implementations, obtaining the adjustment concentrator topology network includes:
[0093] The set of sub-topology structures to be adjusted is eliminated from the multiple concentrator sub-topology structures to obtain a set of fused concentrator sub-topology structures; the first sub-topology structure to be adjusted in the set of sub-topology structures to be adjusted is extracted, and a first neighborhood set adjacent to the first sub-topology structure to be adjusted is matched from the set of fused concentrator sub-topology structures; the concentrators in the first sub-topology structure to be adjusted are randomly adjusted to the first neighborhood set according to a preset adjustment method to obtain a first neighborhood adjustment set and a first adjusted concentrator sub-topology structure, wherein the preset adjustment method is to adjust the concentrators in the first sub-topology structure to be adjusted to the fused concentrator sub-topology structure in the first neighborhood set according to the random adjustment quantity; the neighborhood adjustment optimization is performed on the set of sub-topology structures to be adjusted to obtain multiple neighborhood adjustment sets and multiple adjusted concentrator sub-topology structures; the multiple neighborhood adjustment sets are integrated to obtain multiple neighborhood adjusted concentrator sub-topology structures; the adjusted concentrator topology network is generated according to the multiple adjusted concentrator sub-topology structures and the multiple neighborhood adjusted concentrator sub-topology structures.
[0094] Optionally, a set of sub-topologies to be adjusted is eliminated from the multiple concentrator sub-topologies to obtain a set of concentrator sub-topologies that can be integrated, and the set of concentrator sub-topologies that can be integrated includes multiple neighboring concentrators of multiple topologies in the set of sub-topologies to be adjusted.
[0095] Optionally, extract the first sub-topology to be adjusted, and based on the position relationship and the topological connection relationship, traverse the set of fusible concentrator sub-topology structures to match multiple directly connected concentrators to obtain a first neighborhood set. Wherein, the neighborhood is defined as a concentrator or concentrator topology structure that has a direct connection relationship with the sub-topology to be adjusted in the network topology. Exemplarily, a graph algorithm (such as depth-first search or breadth-first search) is used to find all adjacent nodes of the sub-topology to be adjusted to form a first neighborhood set.
[0096] Optionally, the concentrators in the first sub-topology to be adjusted are randomly adjusted to the first neighborhood set according to a preset adjustment method to obtain a first neighborhood adjustment set and a first adjusted concentrator sub-topology. This includes randomly adjusting the amount (such as by proportion or fixed amount) to move the concentrators in the sub-topology to be adjusted to the first neighborhood set or to move the concentrators in the first neighborhood set to the sub-topology to be adjusted. By randomly adjusting the positions and connection relationships of the concentrators, it is helpful to achieve load balancing and prevent concentrator overload.
[0097] Furthermore, a heuristic optimization algorithm (such as simulated annealing, genetic algorithm) is used to perform neighborhood adjustment optimization on the sub-topology structure set to be adjusted. Through multiple iterative optimizations, the overall fitness of the concentrator sub-topology structure is improved until a satisfactory level is reached, thereby obtaining multiple neighborhood adjustment sets and multiple adjusted concentrator sub-topology structures.
[0098] Optionally, all neighborhood adjustment sets are merged through data integration tools (such as Python's Pandas), and multiple neighborhood adjustment concentrator sub-topology structures are formed after deduplication. Then, based on multiple adjustment concentrator sub-topology structures and multiple neighborhood adjustment concentrator sub-topology structures, multiple concentrator sub-topology structures are updated to generate an adjustment concentrator topology network. Through the above steps, the neighborhood optimization adjustment can be effectively performed on the sub-topology to be adjusted to form an optimized concentrator topology network, thereby improving the overall network performance and resource utilization efficiency.
[0099] Operation management and control is performed based on the adjustment of the concentrator topology network.
[0100] Optionally, based on adjusting the concentrator topology network, the deployment and connection status of the concentrator in the target area is established or updated, and centralized data collection is performed, thereby realizing dynamic update and adaptive adjustment of the concentrator topology in the target area, improving the efficiency of concentrator-based data collection, and enhancing the resource utilization efficiency of the topology network.
[0101] Furthermore, the fitness function is:
[0102] ;
[0103] in, is the topology operation fitness of the Rth concentrator sub-topology corresponding to the Rth operation monitoring data set, m is the number of operation monitoring data in the Rth operation monitoring data set, and m is a positive integer greater than or equal to 1. is the concentrator packet loss rate of the i-th operation monitoring data in the R-th operation monitoring data set, is the concentrator memory utilization of the i-th operation monitoring data in the R-th operation monitoring data set, is the concentrator running speed of the ith running monitoring data in the Rth running monitoring data set, is the content occupancy rate of the server that provides resource support to the Rth concentrator sub-topology.
[0104] Through the above steps, the value of the fitness function is calculated to evaluate the operational fitness of the concentrator subtopology. The fitness function comprehensively considers different indicators of multiple operational monitoring data and combines the content occupancy rate of the server for evaluation. Among them, the operating speed of the concentrator refers to the communication rate between the concentrator and the upper device or server, in bps.
[0105] In summary, the concentrator control method based on adaptive topology provided by the present invention has the following technical effects:
[0106] By extracting multiple uploaded data sets of multiple terminals within a preset window, information entropy calculation is performed on these uploaded data sets to generate multiple information entropies, wherein each terminal has a unique terminal location identifier; the multiple information entropies are respectively used as the heights of multiple peaks of the simulated mountain, and are combined with the multiple terminal location identifiers to input into the simulation model to generate multiple simulated peaks, and the low-lying areas between any adjacent simulated peaks constitute multiple simulated valleys, and an information simulated mountain is constructed based on these simulated valleys and simulated peaks; the simulated valleys corresponding to the lowest points of the valley heights in the multiple simulated valleys are used as watering points, and water is poured into the information simulated mountain, and the information simulated mountain is divided into regions through the rising water level to generate multiple divided regions, and these divided regions include multiple divided information entropy sets; based on these divided information entropy sets, concentrators are configured for the divided regions to obtain multiple concentrator sub-topology structures, and each concentrator sub-topology structure includes multiple concentrator topology nodes; the multiple concentrator sub-topology structures are operated and monitored, and the neighborhood optimization and adjustment of these concentrator sub-topology structures is performed according to the operation monitoring results to form an adjusted concentrator topology network; operation management and control is performed based on the adjusted concentrator topology network. This achieves the technical effect of improving the adaptability of the concentrator and the efficiency of collection and control.
[0107] It should be understood that the embodiments disclosed in the present invention and the above description can enable those skilled in the art to use the present invention to implement the present invention. At the same time, the present invention is not limited to the above-mentioned embodiments. It should be understood that those skilled in the art can still modify the technical solutions recorded in the above-mentioned embodiments, or replace some of the technical features therein by equivalents; and these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention, and should be included in the protection scope of the present invention.
Claims
1. A concentrator control method based on adaptive topology, characterized in that: The method comprises: Extracting multiple uploaded data sets of multiple terminals within a preset window, performing information entropy calculation on the multiple uploaded data sets to generate multiple information entropies, wherein the multiple terminals have multiple terminal location identifiers; The plurality of information entropies are respectively used as the heights of the plurality of peaks of the simulated mountain, and are combined with the plurality of terminal location identifiers to input into the simulation model to generate a plurality of simulated peaks, and the low-lying areas between any adjacent plurality of simulated peaks constitute a plurality of simulated valleys, and the information simulated mountain is constructed based on the plurality of simulated valleys and the plurality of simulated peaks; Taking the simulated valley corresponding to the lowest point of the valley height among the multiple simulated valleys as the watering point, water is poured into the information simulated mountain, and the information simulated mountain is divided into regions according to the rising water level to generate multiple divided regions, wherein the multiple divided regions include multiple divided information entropy sets; Performing concentrator configuration on the multiple divided areas based on the multiple divided information entropy sets to obtain multiple concentrator sub-topology structures, wherein each regional concentrator topology structure includes multiple concentrator topology nodes; Performing operation monitoring on the multiple concentrator sub-topology structures, and performing neighborhood optimization adjustment on the multiple concentrator sub-topology structures according to the operation monitoring results to form an adjusted concentrator topology network; Operation management and control is performed based on the adjustment of the concentrator topology network.
2. The concentrator control method based on adaptive topology according to claim 1, characterized in that: include: Taking the simulated valley corresponding to the lowest point of the valley height among the multiple simulated valleys as the watering point, water is poured into the information simulated mountain, and the first simulated peak that is flooded is taken as the first starting simulated peak; Continue to pour water into the information simulated mountain until the next simulated peak is submerged, thereby obtaining a stage simulated peak; Determine whether the peak height difference between the first starting simulated peak and the stage simulated peak is greater than or equal to a preset difference, and if so, generate a first dividing ridge line at the stage simulated peak, wherein the first dividing ridge line rises as the water level rises; The simulated peak in the stage is used as the second starting simulated peak, and water is continuously poured into the information simulated mountain until the simulated peak corresponding to the maximum peak height is submerged, and the information simulated mountain is divided into multiple divided areas according to multiple dividing ridge lines on the water surface.
3. The concentrator control method based on adaptive topology according to claim 2, characterized in that: include: When the difference in peak height between the first starting simulated peak and the stage simulated peak is less than a preset difference, no dividing ridge line is generated at the stage simulated peak, and water continues to be poured into the information simulated mountain until the simulated peak corresponding to the maximum peak height is submerged, and the information simulated mountain is divided into multiple divided areas according to the multiple dividing ridge lines on the water surface.
4. The concentrator control method based on adaptive topology according to claim 1, characterized in that: include: Traversing and counting the information entropy quantities in the multiple divided information entropy sets to generate multiple information entropy quantities; Performing information entropy superposition on the multiple divided information entropy sets respectively to generate multiple information entropy superposition totals; The concentrator is configured according to the multiple information entropy quantities and the total amount of superposition of the multiple information entropies to obtain the multiple concentrator sub-topology structures.
5. The concentrator control method based on adaptive topology according to claim 4, characterized in that: include: Get a feedforward neural network; Collect n groups of configuration samples, where each group of configuration samples includes multiple sample information entropy quantities, multiple sample information entropy superposition totals, and multiple sample concentrator configuration schemes; The feedforward neural network is trained using the n groups of configuration samples in sequence, and the network parameters of the feedforward neural network are adjusted according to the output results of the previous group of training until convergence, so as to obtain a trained configuration network layer; Using the configuration network layer, the plurality of information entropy quantities and the plurality of information entropy superposition total amounts are respectively identified to generate a plurality of concentrator configuration schemes, wherein each concentrator configuration scheme includes a concentrator configuration quantity, a concentrator layout location, and server allocation resources; Concentrators are arranged in the multiple divided areas according to the multiple concentrator configuration schemes to generate multiple concentrator sub-topology structures.
6. The concentrator control method based on adaptive topology according to claim 1, characterized in that: include: Performing operation monitoring on the multiple concentrator sub-topologies in a preset monitoring window to obtain multiple operation monitoring data sets, wherein each operation monitoring data set includes multiple concentrator operation speeds, multiple concentrator packet loss rates, and multiple concentrator memory utilization rates; Analyzing the plurality of operation monitoring data sets using a fitness function to obtain a plurality of topological structure operation fitnesses; When the operating fitness of the multiple topological structures is less than or equal to the preset fitness, the corresponding concentrator sub-topological structure is added to the sub-topological structure set to be adjusted; The neighborhood adjustment optimization is performed on the set of sub-topology structures to be adjusted to obtain the adjustment concentrator topology network.
7. The concentrator control method based on adaptive topology according to claim 6, characterized in that: include: Eliminating the sub-topology structure set to be adjusted from the multiple concentrator sub-topology structures to obtain a convergible concentrator sub-topology structure set; Extracting a first sub-topology to be adjusted from the set of sub-topology structures to be adjusted, and matching a first neighborhood set adjacent to the first sub-topology to be adjusted from the set of fusible concentrator sub-topology structures; Randomly adjust the concentrators in the first sub-topology to be adjusted to the first neighborhood set according to a preset adjustment method to obtain a first neighborhood adjustment set and a first adjusted concentrator sub-topology, wherein the preset adjustment method is to adjust the concentrators in the first sub-topology to be adjusted to the fused concentrator sub-topology in the first neighborhood set according to the random adjustment quantity; Performing neighborhood adjustment optimization on the sub-topology structure set to be adjusted to obtain multiple neighborhood adjustment sets and multiple adjustment concentrator sub-topology structures; Integrating the plurality of neighborhood adjustment sets to obtain a plurality of neighborhood adjustment concentrator sub-topology structures; The adjustment concentrator topology network is generated according to the plurality of adjustment concentrator sub-topologies and the plurality of neighborhood adjustment concentrator sub-topologies.
8. The concentrator control method based on adaptive topology according to claim 6, characterized in that: The fitness function is: ； in, is the topology operation fitness of the Rth concentrator sub-topology corresponding to the Rth operation monitoring data set, m is the number of operation monitoring data in the Rth operation monitoring data set, and m is a positive integer greater than or equal to 1. is the concentrator packet loss rate of the i-th operation monitoring data in the R-th operation monitoring data set, is the concentrator memory utilization of the i-th operation monitoring data in the R-th operation monitoring data set, is the concentrator running speed of the ith running monitoring data in the Rth running monitoring data set, is the content occupancy rate of the server that provides resource support to the Rth concentrator sub-topology.