A bidirectional grid-connected control system for a low-power energy storage converter

Abstract

The present invention discloses a bidirectional grid-connected control system for a small-power energy storage converter, which relates to the technical field of converter grid-connected control. Aiming at the requirements of a small-power energy storage converter in a limited scattering environment, a communication and control solution is proposed. By arranging measurement points and generating a channel attenuation matrix and an interference feature list through a multi-factor attenuation function, and forming a channel environment model through non-linear fitting; constructing a multi-node multi-band communication topology and screening feasible routes, dynamically solving the optimal or alternative combination and outputting the network topology configuration; performing multi-dimensional scoring and multi-path capacity allocation on data streams with different sensitivities, and preferentially scheduling high-priority data to a better route; when the network anomaly cannot be restored, the local buffer and offline control algorithm can calculate the inverter operation parameters based on the cached data and perform load reduction protection, and smoothly switch back to the online mode after the network is restored, ensuring that the energy storage converter is still safe and controllable in the disconnected state.

Description

Technical Field

[0001] The present invention relates to the technical field of converter grid-connected control, and specifically to a bidirectional grid-connected control system for a small-power energy storage converter. Background Art

[0002] In the context of the rapid development of distributed energy, small-power energy storage converters are increasingly applied to independent power supply or grid-connected systems in households, commercial buildings, and remote areas. Such converters often need to operate in environments such as basements, parking lots, building-dense areas, or agricultural buildings far from the main grid. These scenarios usually have characteristics such as enclosed space, numerous obstacles, and complex radio propagation paths, resulting in severe attenuation, reflection, or multipath interference in communication channels. At the same time, the changes in microgrids or user-side loads in terms of time periods and demands are frequent and unpredictable, putting great pressure on the energy storage converter in terms of power scheduling, grid interaction, and data communication. Without real-time monitoring and adaptive control methods under the combined action of small-power operation and variable environments, the system operation efficiency and safety are easily restricted.

[0003] In the Chinese invention patent with the authorization announcement number CN110266044B, a microgrid grid-connected control system and method based on an energy storage converter are publicly proposed. First, virtual sine and cosine functions are used to quickly detect the negative-sequence current, harmonics, and reactive current of the grid; a proportional-resonant controller is used to achieve power output control and frequency-division compensation control of the reactive power and harmonics of the microgrid grid-connected current according to the system margin, and multi-objective control of the energy storage converter is performed; the compensation and balance of the microgrid grid-connected current can be achieved, and the power quality of the microgrid can be improved. At the same time, due to the complex design of the control parameters of the proportional-resonant controller, a fuzzy algorithm is combined for online tuning of the proportional-resonant parameters, making the system have good dynamic performance.

[0004] Existing grid-connected control schemes for small-power energy storage devices often cannot fully cope with the dual challenges of complex wireless environments and load dynamics. Specifically, when the network quality fluctuates greatly due to scattering, shadow shielding, etc., the converter cannot continuously receive key control instructions from the central dispatcher or remote server, easily leading to safety hazards such as grid-connected power output deviation and instantaneous overshoot of voltage and current. In addition, once the communication link is interrupted, some systems lack edge-side buffering and offline control mechanisms, and it is difficult to make local decisions in a timely manner according to the state of charge of the battery and load changes, resulting in the failure of charge and discharge strategies and causing problems such as battery over-discharge, short-term power outage, or grid instability.

[0005] Therefore, the present invention provides a bidirectional grid-connected control system for a small-power energy storage converter. Summary of the Invention

[0006] (1) Technical Problems to be Solved

[0007] In view of the deficiencies of the prior art, the present invention provides a bidirectional grid-connected control system for a low-power energy storage converter. By addressing the requirements of a low-power energy storage converter in a limited scattering environment, a communication and control solution is proposed. Channel attenuation matrices and interference feature lists are generated through measurement point layout and multi-factor attenuation functions, and a channel environment model is formed through non-linear fitting. A multi-node multi-band communication topology is constructed and feasible routes are screened, and the optimal or alternative combinations are dynamically solved and the network topology configuration is output. Multi-dimensional scoring and multi-path capacity allocation are performed on data streams with different sensitivities, and high-priority data is preferentially scheduled to better routes. When the network anomaly cannot be restored, the local buffer and offline control algorithm can calculate the inverter operation parameters based on the cached data and perform load reduction protection, and smoothly switch back to the online mode after the network is restored, ensuring that the energy storage converter remains safe and controllable in the disconnected state. A technical solution with both communication reliability adaptive strategies and offline control guarantees is proposed to ensure that the low-power energy storage converter can still perform grid-connected control safely and stably and achieve efficient energy management in a limited scattering environment, solving the technical problems raised in the background art.

[0008] (2) Technical solution

[0009] To achieve the above objectives, the present invention is realized through the following technical solutions: A bidirectional grid-connected control system for a low-power energy storage converter includes arranging a number of measurement points in the target deployment area, measuring and recording radio frequency indicators and environmental parameters, integrating the measurement point data into a channel attenuation matrix and an interference feature list through a multi-factor attenuation function and path integration, and performing non-linear fitting and correlation on the integrated data to obtain a channel environment model;

[0010] Generating a communication topology structure of multi-node multi-band, screening feasible routes according to the channel attenuation threshold and incorporating them into the candidate set; constructing a communication quality prediction model and quantitatively analyzing the core indicators of each candidate route, dynamically solving the optimal or alternative route combination, and outputting a network topology configuration plan;

[0011] Performing multi-dimensional scoring on transmission indicators according to the different sensitivities of data categories, and strengthening the distinction between instantaneous and cumulative interference; preferentially injecting high-priority data into better-scored routes and frequency bands through multi-path capacity allocation, and restricting the speed or delaying the transmission of ordinary data;

[0012] Writing the current grid connection command, battery SOC, and operation history data into the local buffer unit, calculating the operation parameters of the inverter by the offline control algorithm based on the local prediction model, performing local load reduction protection on the over-limit state or fault risk and recording alarms, and synchronizing data with the remote end and smoothly switching back to the online mode after the network is restored.

[0013] Further, the distance mark of each measurement point 、measurement point coordinates and the characteristics of nearby obstacles for recording; for each possible operating frequency band, the received signal strength indication is measured in sequence and the number of lost packets , and stored in the initial environment marking table;

[0014] Define a multi-factor attenuation function for the measurement point at the operating frequency band to calculate the comprehensive attenuation amount When calculating the comprehensive attenuation amount for all measurement points and all possible frequency bands one by one , record the result in matrix form as , and at the same time combine the number of lost packets and the received signal strength indication to fuse the interference intensity reference value on the same matrix dimension;

[0015] Refine the key interference feature set according to the obstacle type or spatial distribution, and correspond it with the measurement point coordinates to form an interference feature list.

[0016] Furthermore, use multi-dimensional non-linear fitting to perform multi-parameter fitting on the matrix and the interference feature list to generate a function that can quickly estimate the attenuation value and interference factor for any coordinate and carrier frequency band :

[0017] In the formula represents the function mapping trained by the channel attenuation matrix and interference feature list obtained in step 102 under the machine learning algorithm;

[0018] Package and store the above fitting function in the channel environment model, and retain the parameter index; the finally output channel environment model includes: obstacle interference feature index, reference correction amount corresponding to the frequency band , channel attenuation function , the corresponding relationship between spatial coordinates and measurement point numbers.

[0019] Furthermore, according to the environmental attributes and geographical coordinate information of each measurement point, combined with the actual installation location of the energy storage converter and the optional access point location, generate a set of potential communication nodes , each node in this set contains the corresponding coordinates and the available frequency band set ;

[0020] For any two nodes in the node set and​ , retrieve its attenuation value in the frequency band from the channel environment model ; , if the attenuation value exceeds the tolerable threshold, mark the link as unavailable in the frequency band ; otherwise, temporarily record it as a candidate link and include it in the subsequent routing calculation.

[0021] Furthermore, based on the candidate links, use the graph search method to generate a set of feasible routing sets under multiple available frequency bands ; among them, each route records the sequence of nodes passed through and the available bandwidth and reference delay of the corresponding link;

[0022] Output the candidate route set record, which contains the feasible paths under the available frequency band , and associate the attenuation and interference parameters in the channel environment model with each path.

[0023] Furthermore, to comprehensively evaluate the quality of the candidate routes, define the following multi-dimensional routing weight vector :

[0024] where: represents the estimated packet loss coefficient of route in the frequency band , represents the estimated delay; represents the available bandwidth metric;

[0025] Aggregate the multi-dimensional routing eigenvalue of all candidate paths at the prediction moment or period to construct a communication quality prediction model.

[0026] Furthermore, combined with the grid connection control requirements of the energy storage converter and the potential transmission capacity requirements, set the objective function , select one or more key dimensions in the transmission metrics for optimization, and a composite objective function can be defined:

[0027] where is a weighting coefficient that can be allocated according to the priority;

[0028] Among all the route and frequency band combinations output by the communication quality prediction model, select several paths that minimize as the main candidates, and integrate the selected optimal or alternative paths and their corresponding frequency bands, estimated transmission capacity, priority, etc. information to form a network topology configuration plan.

[0029] Further, the data stream to be transmitted is divided into categories such as critical grid connection control instructions, battery status data, and general monitoring information. Each category is indexed indicated. For the data category under the routing , frequency band conditions, the scoring value is defined by a multi-dimensional scoring function :

[0030] is a three-dimensional index vector, the weighting matrix , dimension , is the weight or penalty matrix designed for the data category , is the upper limit of time integration; is an additional routing / frequency band environment correction term, represents the topological features or environmental markers related to the routing and the frequency band , is a one-dimensional weight vector, is the real-time interference correction amount;

[0031] For the actual transmission requirements of all data categories , they are sorted according to the size of the comprehensive scoring function , and the data categories are mapped to the corresponding priority queues.

[0032] Further, one or more routes suitable for the high-priority data volume are selected from the network topology configuration scheme and the corresponding frequency bands :

[0033] Among them, obtaining the capacity allocation coefficient represents the capacity allocation coefficient for the data category on the route , frequency band , is the current set of all combination candidates; is the sensitive tuning coefficient for the data category ;

[0034] According to the quality of the capacity allocation coefficient , the corresponding traffic ratio is allocated. According to the capacity allocation coefficient calculated for each data category , issue specific scheduling instructions to network nodes.

[0035] Furthermore, during the execution of multipath scheduling, continuously monitor the actual packet loss rate and delay of each active route at present. If it is detected that there is an abnormal increase or the difference from the communication quality prediction model is too large, trigger the immediate switching logic;

[0036] Once it is determined that a switch is needed, according to the multipath capacity allocation alternative plan, quickly output new transmission control instructions and priority settings to the corresponding network nodes.

[0037] Furthermore, if it is found that the actual packet loss rate or delay value of the current route has a significant difference from the predicted value in the communication quality prediction model, and the cumulative exceeds the difference threshold , then trigger a network anomaly alarm; among them, the following formula is used for network anomaly measurement :

[0038] Among them, is the measured packet loss rate, is the predicted value of the packet loss rate by the communication quality prediction model at time , represents the length of the anomaly evaluation interval;

[0039] When , it indicates that this route has a serious anomaly and it is necessary to immediately evaluate whether there are other available routes: if there is no available route, enter the offline mode.

[0040] Furthermore, if there is still no available alternative route after anomaly determination, send an offline mode activation signal to the local buffer unit; notify the control algorithm module to prepare to switch to the offline control strategy so as to stably control key parameters such as the grid-connected power of the energy storage converter and the output voltage / current reference in the case of no remote communication.

[0041] Write the minimum required historical data into the safety buffer area, adopt multi-level version management for key grid-connected parameters, mark the current latest version parameter as , the historical version is marked as , represents the storage order or time slice; use the priority of key grid-connected parameters and ordinary monitoring data to set the capacity allocation of the buffer area.

[0042] Furthermore, load the offline control algorithm to take over the output power, charge and discharge instructions and safety protection logic of the converter; the offline control algorithm can adopt a local prediction model based on the state equation;

[0043] When the battery State of Charge (SOC) approaches the limit threshold or the converter temperature is detected to be too high, the offline control algorithm enables protection measures without remote instructions, including automatically reducing the load or restricting the charge and discharge power. Referring to the internal fault diagnosis table of the system, local records are made for alarms that require manual intervention and reported to the remote platform when the network is restored next time;

[0044] While the offline mode continues to execute, the network is checked for recovery in the minimum power consumption or low-speed polling mode;

[0045] When the network status improves significantly and meets , first, the network link stability is confirmed through the verification mechanism, and then the cached data is synchronously bidirectionally with the remote system.

[0046] (III) Beneficial Effects

[0047] The present invention provides a bidirectional grid-connected control system for a small-power energy storage converter, having the following beneficial effects:

[0048] Through a multi-step and full-process technical solution, the communication reliability of the small-power energy storage converter in a limited scattering environment is deeply combined with the adaptive grid-connected control, effectively making up for the deficiencies of the prior art in complex deployment scenarios. The specific beneficial effects are as follows: Through the comprehensive operation of the multi-factor attenuation function, the environmental impact matrix, and the interference feature list, it is possible to give a refined measurement for multiple scattering, obstacle penetration, and frequency band differences in non-ideal scenarios such as basements and remote farms, forming a "channel environment model" to lay a solid foundation for subsequent network topology design.

[0049] With the help of the network topology configuration and the communication quality prediction model, the optimal or alternative communication path is selected from the indicators such as bandwidth, delay, and packet loss of the candidate routes and frequency bands, enabling the small-power energy storage converter to obtain a highly reliable data interaction path under limited bandwidth and unstable channels. On this basis, the adaptive transmission strategy uses a multi-dimensional scoring function and a capacity allocation formula to distinguish the priorities of grid-connected control instructions, battery state data, and ordinary monitoring information and dynamically allocate bandwidth. Once congestion occurs or the key indicators deteriorate significantly, the routing can be switched immediately to ensure the real-time performance and integrity of the key data.

[0050] When the network has a serious anomaly and cannot be repaired, the local buffering and offline control strategy can perform predictive calculations on the cached grid-connected instructions, battery historical data, etc., ensuring the safety and controllability of the voltage, current, and charge and discharge processes of the energy storage converter even in the situation of temporary disconnection; at the same time, it has the ability of fault detection and automatic load reduction to prevent the battery from over-discharging or overheating and failing. After the network is restored, the system can automatically complete data synchronization and smoothly switch back to the online mode.

[0051] From channel modeling, network topology design to adaptive scheduling and offline control, multiple links progress layer by layer, which not only strengthens the applicability of wireless communication in closed or remote environments, but also ensures the grid connection stability and security of energy storage converters, significantly improving the application scenario adaptability and overall operation efficiency of small power energy storage systems. Description of the Drawings

[0052] Figure 1 This is a schematic structural diagram of the bidirectional grid connection control system for a small power energy storage converter of the present invention. Detailed Implementation Modes

[0053] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0054] Please refer to Figure 1 , the present invention provides a bidirectional grid connection control system for a small power energy storage converter, including

[0055] Step 1: Arrange several measurement points in the target deployment area, measure and record radio frequency indicators and environmental parameters, integrate the measurement point data into the channel attenuation matrix and interference feature list through a multi-factor attenuation function and path integration, and perform non-linear fitting and correlation on the integrated data to obtain a channel environment model;

[0056] Step 101: Arrange several fixed measurement points in the deployment site, and perform initial measurement and marking on the environmental parameters and radio frequency signal indicators at each position according to the preset measurement route and measurement point numbers. The specific technical logic is as follows:

[0057] Use mobile or fixed acquisition equipment to mark the distance of each measurement point , measurement point coordinates and the characteristics of nearby obstacles for recording; for each possible working frequency band (such as a specific frequency band of 24 GHz, 5 GHz or LoRa, etc.), sequentially measure the received signal strength indication and the number of lost packets , and at the same time record the acquisition time for subsequent association with time variables;

[0058] Store the measurement point numbers, coordinate information, obstacle types and radio frequency measurement values in the initial environment marking table, and complete the preliminary measurement and marking of environmental parameters and radio frequency indicators for all measurement points;

[0059] During use, through systematic measurement of the distance, coordinates of measurement points, and obstacle types, the true situation of the deployment area in terms of spatial distribution, terrain structure, and potential shielding factors can be fully understood, providing an accurate basis for subsequent model construction; successively collecting and recording the signal strength, packet loss number, etc. of possible frequency bands such as 2.4 GHz, 5 GHz, and LoRa enables the system to multi-dimensionally evaluate the feasibility or advantages and disadvantages of different frequency bands in subsequent designs, taking into account the characteristics of multiple frequency bands; uniformly storing the numbers of these measurement points, obstacle characteristics, and real-time acquisition results in the initial environment marking table helps ensure data consistency, support the scalability of subsequent algorithm analysis, and achieve unified data management;

[0060] Step 102: Integrate and calculate the channel attenuation and interference conditions of each measurement point by establishing a multi-factor attenuation function to generate a channel attenuation matrix and extract interference characteristics. The core technical logic is as follows:

[0061] Define the multi-factor attenuation function for the measurement point at the working frequency band to calculate the comprehensive attenuation amount when introducing the combined form of the environmental impact matrix and path integration, which is defined as follows:

[0062] In the formula: represents the comprehensive attenuation amount of the measurement point at the frequency band and is used to make difference comparisons or fusions with the measured received signal strength indication and the packet loss number to form the final channel attenuation matrix and interference characteristic list. The larger its value, the more serious the attenuation of the wireless signal;

[0063] The environmental impact matrix E discretizes the overall test scenario into several grids or blocks (such as coordinate grids), and defines an environmental impact vector (including humidity, temperature, wall material, scattering coefficient, etc.) for each grid. These vectors form the environmental impact matrix E after combination. When it is necessary to query the environment where the measurement point is located, the measurement point environment vector can be indexed from the environmental impact matrix , that is, the environmental impact vector corresponding to the measurement point .

[0064] The frequency band weight vector represents the weighting coefficient of each component of the environmental impact vector at the frequency band and is used to weight the measurement point environment vector After performing the dot product of vectors, the basic attenuation of the measurement point at the corresponding frequency band can be obtained, which is calibrated by machine learning or experience in the early stage and may be adaptively adjusted for different regions in the future;

[0065] Measurement point environment vector , the measurement point extracted from the environmental impact matrix The corresponding environmental attribute vector, such as humidity, temperature, scattering coefficient; reflecting the static or semi-static environmental factors of the area where the measurement point is located.

[0066] Path set , which refers to the set of discrete paths from the emission source (or reference source) to the measurement point The path may pass through several grids or regions and can be determined by the shortest path algorithm or a multi-path model based on beam directivity / reflection / diffraction;

[0067] is the variable along the path in the integral, usually understood as distance or some path parameter, which varies with the path discretization;

[0068] is the instantaneous attenuation coefficient on the integral segment of the path, indicating the type of medium or obstacle at the path position ; this function can be comprehensively determined by the refraction / reflection characteristics of local obstacles, material penetration loss, and the attenuation law of the frequency band The path integral term is used to accumulate the local attenuation effect, making the modeling of complex scenarios (multiple reflections, wall penetration, diffraction, etc.) more accurate;

[0069] represents the cancellation or compensation term related to the main obstacle type around the measurement point When there are large-scale or special material obstacles at the measurement point , the additional influence of this area on attenuation can be reflected by increasing the value of ; in actual calculations, it can be positive or negative;

[0070] Calculate the comprehensive attenuation for all measurement points and all possible frequency bands one by one, and record the results in matrix form as . At the same time, combine the packet loss number and the received signal strength indication to fuse the interference intensity reference value on the same matrix dimension; after the matrix is calculated, further refine the key interference feature set according to the obstacle type or spatial distribution, such as extreme loading positions, high-interference regions, etc., and combine it with the measurement point coordinates​​ Correspond to form a list of interference features;

[0071] When in use, through the combination of the environmental impact matrix and path integral, more refined cumulative modeling can be carried out for complex scattering, reflection and other factors, avoiding the errors brought by the traditional simple distance model. After calculating the channel attenuation matrix, aggregate analysis is carried out on the measurement points that may have extreme attenuation or high interference to form a list of interference features, providing a direct reference for risk control in network planning.

[0072] Step 103: Based on the channel attenuation matrix and the list of interference features, perform deeper fitting and model solidification, and output the final channel environment model for subsequent network topology design and communication quality prediction. The specific technical logic is as follows: Adopt a multi-dimensional non-linear fitting or machine learning algorithm (such as Gaussian process regression based on kernel functions, etc.) to perform multi-parameter fitting on the matrix and the list of interference features to generate a function that can quickly estimate the attenuation value and interference factor for any coordinate and carrier frequency band :

[0073] In the formula represents the function mapping trained by the channel attenuation matrix and the list of interference features obtained in step 102 under the machine learning algorithm;

[0074] Encapsulate and store the above fitting function in the channel environment model, and retain the parameter indexes such as frequency band, coordinate, and obstacle type respectively, so that the corresponding attenuation and interference estimation values can be quickly obtained when querying any coordinate and frequency band in the subsequent steps; The finally output channel environment model includes: obstacle interference feature index, reference correction amount corresponding to the frequency band , channel attenuation function , the corresponding relationship between spatial coordinates and measurement point numbers;

[0075] When in use, encapsulate the fitted attenuation function and interference information in the channel environment model, allowing quick obtaining of attenuation and interference estimation values under any coordinate and frequency band combination, greatly reducing the complexity of subsequent topology design and prediction operations. Through the training and solidification of interference features and multi-dimensional attenuation matrices, the model can be quickly updated when extended to new coordinates, adding or reducing obstacles, or adding new frequency bands, with good versatility and scalability, and can flexibly adapt to new scenarios.

[0076] Step 2: Generate a communication topology structure with multiple nodes and multiple frequency bands, screen feasible routes according to the channel attenuation threshold and include them in the candidate set; construct a communication quality prediction model and quantitatively analyze the core indicators of each candidate route, dynamically solve the optimal or alternative route combination, and output the network topology configuration plan;

[0077] Step 201: Based on the environmental attributes and geographical coordinate information of each measurement point in the channel environment model obtained in Step 1, combined with the actual installation location of the energy storage converter, the optional access point locations, etc., generate a set of potential communication node sets , and each node in this set contains the corresponding coordinates and the available frequency band set ;

[0078] For any two nodes in the node set and , retrieve their attenuation values at the frequency band from the channel environment model. If the attenuation value exceeds the tolerable threshold, mark the link as unavailable at the frequency band ; otherwise, temporarily record it as a candidate link and include it in the subsequent routing calculation; Based on the candidate links, use a graph search method (such as Dijkstra, A*, or a custom multi-weight search) to generate a set of feasible routing sets for each pair of key nodes (such as the energy storage converter device node and the gateway node) at multiple available frequency bands

[0079] ; among them, each route records the sequence of nodes passed through and the available bandwidth, reference delay, etc. of the corresponding links, providing input for the next communication quality prediction; output a candidate route set record, which contains the feasible paths at the available frequency band , and associate the attenuation and interference parameters in the channel environment model with each path; When in use, abstract the deployment location of the energy storage converter, the optional access point locations, etc. as nodes, and combined with the channel environment model in Step 1, it can quickly determine which frequency bands are feasible for communication between nodes, thereby clarifying the communicable nodes and frequency bands; by retrieving the attenuation values between candidate nodes, automatically filter out links with excessive attenuation or exceeding the system threshold, ensuring that the subsequent routing search is only for actually available links, reducing invalid calculations, and eliminating unavailable links;

[0080] Step 202: To comprehensively evaluate the quality of candidate routes, define the following multi-dimensional routing weight vector

[0081] : :

[0082] Wherein: represents the estimated packet loss coefficient of the route at the frequency band , represents the estimated delay; represents the available bandwidth metric;

[0083] To achieve high-precision modeling of packet loss and latency, a prediction formula based on integration and environmental cumulative effects is introduced to obtain the estimated packet loss rate. :

[0084] In the formula: represents the candidate path, represents adjacent node pairs in the path; is from node to node the spatial distance is the environmental feature vector at position on the path, which can be indexed from the channel environment model;

[0085] is the instantaneous coefficient of cumulative packet loss attenuation in the frequency band to quantify the additional interference caused by the medium or obstacles to the signal; is the correction factor introduced considering multiple factors such as node handover overhead and potential collision domain diffusion;

[0086] Thus, the packet loss estimates for different links between nodes can be accumulated segment by segment to obtain the overall packet loss estimate of the route; the corresponding latency can also be calculated by a similar principle in combination with the transmission rate and processing duration, and fine-tuned based on the real-time / offline parameters provided by the channel environment model;

[0087] Based on the available bandwidth estimate of the candidate path, considering the concurrent transmission and interference conditions of other nodes in the network, the available bandwidth of each link is corrected twice to form the final available bandwidth index ; The bandwidth dynamics of multiple paths under different load conditions can be estimated using a collaborative prediction or bandwidth competition model, and the results are also stored synchronously in the sub-module of the communication quality prediction model;

[0088] Summarize the multi-dimensional routing eigenvalue of all candidate paths at the prediction moment or period to construct a communication quality prediction model, covering multi-dimensional indicators such as packet loss, latency, and bandwidth in different frequency bands, paths, and time windows, and the model can be trained with time series as needed (for example, using deep neural networks or recursive models);

[0089] Using comprehensive metrics such as packet loss, latency, and bandwidth to help the system make quantitative comparisons among multiple candidate paths, rather than just looking at a single metric, improves the refinement of network deployment and enables multi-dimensional routing metric evaluation. Introducing an integral calculation of obstacles and medium characteristics on the path segment can more realistically present the propagation loss of wireless signals in a local scenario, making the prediction results closer to reality. Incorporating the evaluation of concurrent transmissions from other nodes further flexibly corrects the bandwidth resources during peak and idle loads, avoiding blindly overestimating or underestimating the available bandwidth of a certain path. Serializing the packet loss, latency, and bandwidth of each route at different times and environmental states can be combined with machine learning or deep neural networks for training, thus enabling a certain degree of self-learning ability.

[0090] Step 203: Set the objective function in combination with the grid connection control requirements of the energy storage converter and the potential transmission capacity requirements. , select one or more key dimensions for optimization among transmission metrics such as packet loss, latency, and bandwidth, and a composite objective function can be defined:

[0091] where is a weighting coefficient that can be allocated according to priority;

[0092] Among all the route and frequency band combinations output by the communication quality prediction model, select several paths that minimize (or maximize the bandwidth part) of as the main candidates, and they can be dynamically allocated according to different time periods. For alternative paths with no obvious differences, list them together for switching reserves during subsequent adaptive scheduling.

[0093] Integrate the selected optimal or alternative paths and their corresponding frequency bands, estimated transmission capacity, priority, etc. information to form a network topology configuration plan. This plan includes: the available frequency bands and rated throughput of the physical or logical links between each node, the preferred recommended routes under different time windows or environmental states; the estimated packet loss and latency thresholds for each route;

[0094] When in use, by performing weighted combinations of multi-dimensional metrics such as packet loss, latency, and bandwidth, it adapts to the specific requirements of the system in different application scenarios (such as emphasizing high bandwidth, low latency, or high reliability, etc.), quickly selects several optimal paths from all candidate combinations, and performs priority marking; during network operation, it can switch or perform load distribution accordingly at any time to improve communication reliability and efficiency and determine the optimal or alternative routes; integrate key information such as frequency bands, rated throughput, and packet loss thresholds in the same configuration plan to ensure that the system can directly reference them when implementing subsequent transmission strategies, significantly shortening the response time of deployment scheduling.

[0095] Step 3: Perform multi-dimensional scoring on transmission metrics according to the different sensitivities of data categories, and strengthen the distinction between instantaneous and cumulative interference; through multi-path capacity allocation, inject high-priority data into the routes and frequency bands with better scores first, limit the speed or postpone the transmission of ordinary data, and trigger the switching of alternative routes in real time if the performance of the main route continues to deteriorate;

[0096] Step 301: Divide the data stream to be transmitted into categories such as key grid connection control instructions, battery status data, and ordinary monitoring information, and each category is indexed to indicate that the required target route performance requirements in the communication quality prediction model in Step 2 are also different.

[0097] Combined with the packet loss metric , delay metric and bandwidth metric , and add the link status correction term obtained from real-time monitoring (such as the attenuation correction caused by instantaneous interference),

[0098] For the data category in the route , frequency band Under the conditions, define the following multi-dimensional scoring function :

[0099] Where: is a three-dimensional index vector that depicts the network metrics that this solution is most concerned about at time :

[0100] Among them, the packet loss metric represents the measure of the actual packet loss rate or packet loss coefficient of the route , frequency band at time , the delay metric represents the round-trip delay or delay characteristics of the route , frequency band , and the available bandwidth metric can represent the available bandwidth or throughput metric at time ; The available bandwidth or throughput metric at time

[0101] Weight matrix , dimension , is the weight or penalty matrix designed for the data category to reflect the interaction effects between packet loss, delay and bandwidth. For example, if the key grid connection instruction is extremely sensitive to both packet loss and delay, the weight can be increased at the corresponding position in to make the scoring function more sensitive to the changes of these two metrics; generally, it is required in the numerical setting is positive definite or positive semi - definite to ensure that the integral produces a positive penalty for negative factors (such as packet loss) and can produce incentives or reverse inhibition for positive factors (such as high bandwidth);

[0102] is the upper limit of time integral, representing the time window length concerned when evaluating the network quality for data categories It can be designed according to the real - time nature or fault tolerance of the data. Key control instructions may only require a very short window to ensure fast response, while general monitoring information can use a longer window for comprehensive evaluation;

[0103] is an additional routing / band environment correction term, indicating the topological features or environmental markers related to routing and bands such as the number of nodes, coverage range, available access points, etc., which can be extracted from the network topology configuration scheme or the channel environment model, is a one - dimensional weight vector. Through the mapping calculation with , it reflects the advantages or disadvantages of a specific path or band in space. For example, it can increase the score to prefer a more stable or better - covered band; is the real - time interference correction amount. When it is detected that the network has instantaneous congestion or external strong interference resulting in packet loss and large fluctuations in delay, rapid correction can be made here, and this item is superimposed on the integral result to achieve a sensitive response to unexpected fluctuations.

[0104] For the actual transmission requirements of all data categories , sort them according to the size of the comprehensive scoring function . The higher the score, the higher the sensitivity of the data of this category to network quality, or the worse the selected combination. Therefore, higher - priority guarantee is required (or a better route and band need to be found), and the data categories can be mapped to the corresponding priority queues;

[0105] When in use, data categories such as grid - connection control instructions, battery status data, and general monitoring information can be distinguished, with different sensitivities to the three major indicators of packet loss, delay, and bandwidth, so that key data can obtain higher - level protection. By adding a real - time interference correction term to the scoring function, the difference between the communication quality prediction model and the current actual situation can be taken into account to prevent the lag caused by relying only on prediction. If a scoring method based on time integral and matrix weighting is adopted, the interaction effects between indicators such as packet loss and delay can be fully incorporated, making the scoring result more reliable and accurate.

[0106] Step 302: According to the priority classification result obtained in Step 301, select one or more routes from the network topology configuration scheme generated in Step 2 that are adapted to the high-priority data volume and the corresponding frequency bands , meanwhile, for ordinary monitoring data or secondary-priority data, other different routes can be selected to reduce the interference to high-priority data;

[0107] To allocate data load among multiple feasible paths and reduce the risks of packet loss and delay, the following multi-path capacity allocation formula is introduced to obtain the capacity allocation coefficient :

[0108] Among them, obtaining the capacity allocation coefficient represents the capacity allocation coefficient for data category on route , frequency band , and comes from the comprehensive scoring function defined in Step 301;

[0109] is the sensitive tuning coefficient for data category , which is used to amplify or reduce the impact of scoring differences on capacity allocation;

[0110] According to the quality of the capacity allocation coefficient , allocate the corresponding traffic proportion. The smaller the scoring value (the better the network quality), the larger the capacity allocation coefficient; according to the capacity allocation coefficient calculated for each data category , send specific scheduling instructions to network nodes, including: which data streams take which routes during which time periods, the allocation proportion when using which frequency band to send key data in multi-path parallel, and the rate-limiting or delay-sending strategy for ordinary monitoring data, so as not to occupy too much bandwidth.

[0111] When in use, make full use of network redundancy. By sending or allocating data streams in parallel on different paths, it can quickly divert when congestion or attenuation increases on any path, improving the success rate of high-priority data. Based on the scoring difference for exponential normalization allocation, automatically direct the traffic to the routes and frequency bands with better scores, reducing the impact of the main route bottleneck on the overall system transmission, and achieving adaptive capacity allocation; not only giving priority to ensuring the real-time performance of key data, but also reasonably arranging the sending of ordinary monitoring data on the remaining bandwidth, making the overall bandwidth utilization rate higher and reducing unnecessary delays.

[0112] Step 303: Continuously monitor each active route currently during the execution of multi-path scheduling The actual packet loss rate and latency. If an abnormal increase is detected or the difference from the communication quality prediction model in Step 2 is too large, the immediate switching logic is triggered;

[0113] Introduce the following switching judgment function:

[0114] Where: is the measured packet loss index at the current moment, is the measured latency index at the current moment;

[0115] is a non - linear gain function, used to amplify the latency impact after exceeding a certain critical value, so as to trigger switching more quickly; if exceeds the preset threshold , it is determined that the performance of this route has deteriorated, and it should be immediately switched to the next high - quality route selected in Step 302.

[0116] Once it is determined that a switch is needed, according to the multi - path capacity allocation alternative plan left in Step 302, quickly output new transmission control instructions and priority settings to the corresponding network nodes, where: stop sending high - priority data on the current failed or deteriorated route; immediately enable the alternative route and allocate capacity on this route according to the calculated above; resynchronize the sending queues of key grid - connection control instructions and battery status data.

[0117] During use, synchronously monitor the measured values such as packet loss and latency and the prediction differences. Once the deterioration degree exceeds the threshold, immediately switch the alternative route to avoid critical data being at high packet - loss risk for a long time, and achieve rapid detection and response to route deterioration; through the multi - path allocation plan pre - stored in Step 302, it is possible to quickly connect to the standby route, ensure that the interruption of critical data during the switching process is minimized, and complete seamless switching; immediately synchronize the sending queues of grid - connection instructions and battery data after switching, achieve transparent switching for upper - layer applications, and do not affect the logical continuity of the energy storage converter grid - connection control.

[0118] Step Four: Write the current grid - connection instruction, battery SOC, and operation history data into the local buffer unit. The offline control algorithm calculates the operation parameters of the inverter based on the local prediction model, performs local load reduction protection on the over - limit state or fault risk and records the alarm. After the network is restored, synchronize the data with the remote end and smoothly switch back to the online mode;

[0119] Step 401: Continuously receive the link monitoring results output by Step 3. If it is found that the measured packet loss rate or delay value of the current route has a significant difference from the predicted value in the communication quality prediction model, and the cumulative exceeds the difference threshold , then trigger a network anomaly alarm;

[0120] The network anomaly metric is calculated using the following formula :

[0121] where is the measured packet loss rate, is the predicted value of the packet loss rate by the communication quality prediction model at time and represents the length of the anomaly evaluation interval;

[0122] When it indicates that the route has a serious anomaly and it is necessary to immediately evaluate whether there are other available feasible routes: if there is no feasible route, enter the offline mode;

[0123] If there is still no available alternative route after the above anomaly determination (or the alternative route is also in an abnormal state during the same time period), an offline mode activation signal will be sent to the local buffer unit; notify the control algorithm module to prepare to switch to the offline control strategy so that key parameters such as the grid-connected power of the energy storage converter and the output voltage / current reference can be stably controlled even without remote communication.

[0124] When in use, the difference between the measured packet loss and the predicted value is evaluated in integral form, which can better distinguish between persistent degradation and instantaneous fluctuations and reduce unnecessary offline switching frequencies. If there is no available feasible route, the offline mode is automatically activated and an alarm signal is sent to ensure that the energy storage converter will not be in an unprotected dangerous state when the network is completely disconnected.

[0125] Step 402: Receive the latest batch of grid connection control instructions, battery state data (including SOC, SOH) and historical operation information retained in step three and encapsulate and archive them in the local buffer unit; write the minimum required historical data (such as charge and discharge records and voltage and current reference curves in the last few minutes or hours) into the safety buffer so that the subsequent offline control algorithm can infer the real-time working state of the converter based on this data;

[0126] Adopt multi-level version management for key grid connection parameters, mark the current latest version parameter as and the historical version as ( represents the storage order or time slice), ensuring that any damaged or incomplete version can be replaced by an earlier version;

[0127] Introduce hash verification or related encryption verification means to generate verification information for each batch of cached data, preventing errors in off-line control caused by data corruption during accidental power outages or cache failures; utilize the priorities of key grid-connected parameters and ordinary monitoring data to set the capacity allocation of the buffer: critical data enjoys higher storage priority and longer retention time. When the space of the local buffer unit becomes tense, some historical ordinary monitoring information is automatically eliminated, and only necessary grid-connected parameters and battery health data are retained.

[0128] During use, through version management and hash verification, it can effectively prevent incorrect control caused by accidental cache damage, ensuring that the energy storage converter can rely on the latest batch of complete and correct data in the off-line mode, and realizing multi-level version protection of critical data; the priority-driven cache allocation can preferentially retain key grid-connected parameters and battery health information when the space is insufficient, taking into account both system safety and storage efficiency. Prepare the necessary historical operation data and grid-connected instruction versions for the off-line control algorithm, and once switched to off-line, it can enter autonomous control without additional waiting.

[0129] Step 403: When it is determined in Step 401 to enter the off-line mode and the cache operation in Step 402 is completed, load the off-line control algorithm to take over the output power, charge and discharge instructions, and safety protection logic of the converter; the off-line control algorithm uses a local prediction model based on the state equation to autonomously calculate the reference trajectories of the voltage and current ( :

[0130] In the formula: is the system matrix of the local model, which is dynamically adjusted according to the key grid-connected parameters and battery state data stored in the early stage, is the control gain mapping matrix; represents the correction amount given by the off-line control algorithm at time to keep the power, frequency or voltage within the safe range.

[0131] When the battery SOC is close to the limit threshold, or when it is detected that the converter temperature is too high, the off-line control algorithm enables protection measures without remote instructions, including automatically reducing the load or restricting the charge and discharge power. Refer to the internal fault diagnosis table of the system, locally record the alarms that require manual intervention, and report them to the remote platform when the network is restored next time;

[0132] While the off-line mode is continuously executed, check whether the network is restored in the minimum power consumption or low-speed polling mode;

[0133] When the network status is significantly improved and meets , first confirm the stability of the network link through the verification mechanism and then synchronize the cached data with the remote system bidirectionally;

[0134] After synchronization is completed, the offline control algorithm is gradually deactivated, and the control right is returned to the adaptive transmission strategy and real-time scheduling. The new grid connection instructions and network policies are updated to the converter control loop again, and finally the online mode is restored.

[0135] Through the state equation and rolling prediction, the energy storage converter can still finely adjust the voltage, current and power output without remote instructions, avoiding crude strategies such as simple one-size-fits-all or shutdown. When the battery or temperature exceeds the limit, the local algorithm can automatically reduce the load, limit the current and record the alarm to maintain the safety bottom line of the system. After the network is restored, it will be reported uniformly. After the network is restored, first confirm the communication stability, then synchronize all data with the remote end, and finally transfer the control right back to the online adaptive scheduling in an orderly manner to avoid secondary impacts caused by abrupt switching.

[0136] Those of ordinary skill in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed in hardware or software depends on the specific application and design constraints of the technical solution. Professional technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of this application.

[0137] Those skilled in the art can clearly understand that for the convenience and simplicity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the foregoing method embodiments, and will not be described herein again.

[0138] In several embodiments provided in this application, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is only for some logical function divisions. In actual implementation, there may be other division methods. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the displayed or discussed mutual coupling or direct coupling or communication connection can be through some interfaces. The indirect coupling or communication connection of the devices or units can be in electrical, mechanical or other forms.

[0139] The units described as separate components may or may not be physically separated. The components displayed as units may or may not be physical units, that is, they can be located in one place, or they can be distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0140] As described above, it is only the specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art within the technical scope disclosed by the present application can easily think of changes or substitutions, which should all be covered within the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims described above.

Claims

1. A bidirectional grid-connected control system for small power energy storage converters, characterized in that: include, Several measurement points are deployed in the target deployment area to measure and record RF indicators and environmental parameters. The measurement point data are integrated into the channel attenuation matrix and interference feature list through multi-factor attenuation function and path integral. The integrated data is nonlinearly fitted and correlated to obtain the channel environment model. Generate a multi-node and multi-band communication topology structure, screen feasible routes according to the channel attenuation threshold and include them in the candidate set; build a communication quality prediction model and quantitatively analyze the core indicators of each candidate route, dynamically solve the optimal or alternative route combination, and output the network topology configuration plan; The transmission indicators are scored in multiple dimensions according to the different sensitivities of data categories, and the distinction between instantaneous and cumulative interference is strengthened. High-priority data is injected into routes and frequency bands with better scores through multi-path capacity allocation, and ordinary data is limited or sent later. The current grid-connected instructions, battery SOC and operating history data are written into the local buffer unit. The offline control algorithm calculates the operating parameters of the inverter based on the local prediction model. Local load reduction protection is performed for over-limit conditions or fault risks and alarms are recorded. After the network is restored, data is synchronized with the remote end and smoothly switched back to online mode.

2. The bidirectional grid-connected control system for a small power energy storage converter according to claim 1, characterized in that: Distance mark for each measuring point , measuring point coordinates and nearby obstacle characteristics Record; for each possible working frequency band, measure the received signal strength indicator in turn and packet loss , stored in the initial environment tag table; Define a multi-factor attenuation function for the measurement point In the working frequency band The comprehensive attenuation is calculated as When all measurement points and all possible frequency bands Calculate the comprehensive attenuation one by one , and record the result in matrix form as , combined with the number of packet losses and received signal strength indicator , fuse the interference intensity reference value in the same matrix dimension; Extract key interference feature sets based on obstacle type or spatial distribution and compare them with the measurement point coordinates. Correspondingly, a list of interference features is formed.

3. The bidirectional grid-connected control system for a small power energy storage converter according to claim 2, characterized in that: Based on multidimensional nonlinear fitting matrix And interference feature list for multi-parameter fitting, generate any coordinate and carrier frequency band Quickly estimate the attenuation value as a function of the interference factor: In the formula Represents the function mapping between the channel attenuation matrix and the interference feature list trained under the machine learning algorithm; is the obstacle type mapping function, used to characterize the coordinates Possible obstacles Additional attenuation, scattering or reflection effects on signal propagation; The fitting function is encapsulated and stored in the channel environment model, and the parameter index is retained; the output channel environment model includes: obstacle interference feature index, reference correction value corresponding to the frequency band , channel attenuation function , the correspondence between spatial coordinates and measuring point numbers.

4. The bidirectional grid-connected control system for a small power energy storage converter according to claim 3, characterized in that: According to the environmental attributes and geographic coordinate information of each measuring point, combined with the actual installation location of the energy storage converter and the location of the optional access point, a potential communication node set is generated. , each node in the set All contain corresponding coordinates and available frequency bands ; For node collection Any two nodes and , retrieve its frequency band from the channel environment model The attenuation value under If the attenuation value exceeds the tolerable threshold, the corresponding link will be Otherwise, it is temporarily recorded as a candidate link and included in subsequent routing calculations.

5. The bidirectional grid-connected control system for a small power energy storage converter according to claim 4, characterized in that: Based on the candidate links, a graph search method is used to generate a set of feasible routing sets in multiple available frequency bands. ; Each route Record the node sequence passed and the available bandwidth and reference delay of the corresponding link; Output candidate route set records, which contain available frequency bands The feasible paths under the condition are obtained, and the attenuation and interference parameters in the channel environment model are associated with each path.

6. The bidirectional grid-connected control system for a small power energy storage converter according to claim 5, characterized in that: In order to comprehensively evaluate the quality of candidate routes, the following multidimensional routing weight vector is defined: : in: Indicates the route In the frequency band The estimated packet loss coefficient under Indicates the estimated delay; Indicates the available bandwidth indicator; All candidate paths Multidimensional routing feature values ​​at the predicted time or time period The data are summarized and constructed into a communication quality prediction model.

7. The bidirectional grid-connected control system for a small power energy storage converter according to claim 6, characterized in that: Combined with the grid-connected control requirements of the energy storage converter and the potential transmission capacity requirements, the objective function is set , select one or more key dimensions in the transmission index for optimization, and define a composite objective function: in is a weighting factor that can be allocated according to priority; Among all the route and frequency band combinations output by the communication quality prediction model, select The smallest several paths are taken as the main candidates, and the selected optimal or alternative paths and their corresponding frequency bands, estimated transmission capacity, and priority information are integrated to form a network topology configuration plan.

8. The bidirectional grid-connected control system for a small power energy storage converter according to claim 7, characterized in that: The data stream to be transmitted is divided into key grid-connected control instructions, battery status data, and general monitoring information. Each category is indexed Indicates that for data categories In routing , frequency band The scoring value under the condition defines the multi-dimensional scoring function : is a three-dimensional index vector, a weighted matrix , dimension , is the data category The designed weight or penalty matrix, is the upper limit of time integral; For additional routing / band environment correction items, Representation and Routing and frequency band Related topological features or environmental markers, is a one-dimensional weight vector, is the real-time interference correction; For all data categories The actual transmission demand is calculated according to the comprehensive scoring function. Sort by size and map the data categories to the corresponding priority queues.

9. The bidirectional grid-connected control system for a small power energy storage converter according to claim 8, characterized in that: Select one or more routes from the network topology configuration that are suitable for the high-priority data volume And the corresponding frequency band In order to distribute data load among multiple feasible paths and reduce the risk of packet loss and delay, the following multi-path capacity allocation formula is introduced to obtain the capacity allocation coefficient : Among them, obtain the capacity allocation coefficient Indicates the data category In routing , Frequency Band The capacity allocation coefficient on For all currently available options Combine candidate sets; For data category Sensitivity tuning coefficient of Based on capacity allocation factor The quality of each data category is determined by the corresponding traffic ratio. Calculated capacity allocation factor , and issue specific scheduling instructions to network nodes.

10. The bidirectional grid-connected control system for a small power energy storage converter according to claim 9, characterized in that: Continuously monitor each active route during multipath scheduling The actual packet loss rate and delay are detected. If an abnormal increase is detected or the difference with the communication quality prediction model is too large, the instant switching logic is triggered; Once it is determined that switching is required, new transmission control instructions and priority settings are quickly output to the corresponding network nodes based on the multi-path capacity allocation alternative plan.

11. The bidirectional grid-connected control system for a small power energy storage converter according to claim 10, characterized in that: If the current route is found The measured packet loss rate or delay value is significantly different from the predicted value in the communication quality prediction model, and the cumulative difference exceeds the difference threshold. , then a network anomaly alarm is triggered; the following formula is used to measure network anomaly : in, is the measured packet loss rate, For the communication quality prediction model at time The predicted value of packet loss rate, Indicates the length of the abnormal evaluation interval; when When A serious exception occurs and it is necessary to immediately evaluate whether there are other feasible routes available: if there is no feasible route, enter offline mode.

12. The bidirectional grid-connected control system for a small power energy storage converter according to claim 11, characterized in that: If there is still no available alternative route after abnormal determination, an offline mode activation signal will be sent to the local buffer unit; the control algorithm module will be notified to prepare to switch to the offline control strategy so that the grid-connected power and output voltage / current reference of the energy storage inverter can be stably controlled without remote communication; The minimum required historical data is written into the security cache area, and multi-level version management is adopted for key grid-connected parameters. The current latest version parameters are marked as , historical versions are marked as , Indicates the storage order or time slice; uses the priority of key grid-connected parameters and common monitoring data to set the capacity allocation of the cache area.

13. The bidirectional grid-connected control system for a small power energy storage converter according to claim 12, characterized in that: Loading the offline control algorithm takes over the output power, charge and discharge instructions and safety protection logic of the converter; the offline control algorithm can adopt a local prediction model based on the state equation; When the battery SOC approaches the limit threshold, or the converter temperature is detected to be too high, the offline control algorithm activates protection measures without remote instructions, including automatic load reduction or limiting the charge and discharge power. Refer to the fault diagnosis table inside the system, make local records of alarms that require manual intervention, and report them to the remote platform when the network is restored next time; While the offline mode continues to execute, check whether the network is restored by using minimum power consumption or low-speed polling; When the network status improves significantly and satisfies ,First, confirm that the network link is stable through a verification mechanism, and then the cached data is bidirectionally synchronized with the remote system.

Images