A building multi-stage power distribution collaborative prediction system and control method of coupling networking

Abstract

The application discloses a kind of coupled networking's building multi-level power distribution collaborative prediction system and control method, the present application relies on perception layer, algorithm control layer, three-layer decoupling layered architecture, constructs the dual-mode coupling channel transmission model based on power carrier wire, establishes "end load feature identification-mid-level situation aggregation research and judgment-top-level global steady-state optimization" progressive prediction quantification system, realizes three-level distribution box according to end load electrical evolution characteristics pre-action, two-level distribution box according to three-level box aggregation situation gradient prediction, one-level distribution box according to global level coupling situation global regulation, change traditional power distribution "fault post passive protection" shallow control logic, construct multi-level building power distribution network active prediction, progressive prevention and control, global cooperation, adaptive iteration New intelligent quantification control system.

Description

Technical Field

[0001] This invention relates to the fields of intelligent networking, hierarchical collaborative protection and predictive control technology for low-voltage power distribution in buildings, and particularly to a coupled networked multi-level power distribution collaborative predictive system and control method for buildings. Background Technology

[0002] Modern building low-voltage power distribution systems widely adopt a power supply and distribution pattern of "primary main distribution box – secondary distribution box – tertiary terminal distribution box". There is strong electrical coupling between upper and lower level distribution units, load timing correlation, and fault propagation characteristics. Different terminal power units are carried at each level, including resistive circuits for lighting, inductive circuits for air conditioning and other equipment, and stray load circuits for sockets. Traditional building power distribution protection systems, based on local threshold passive response control strategies, have core defects: isolated control logic, lagging situational awareness, and lack of hierarchical computational support. Existing power distribution protection technologies all use a delayed control mechanism of "passive tripping after an electrical fault exceeds the limit," failing to establish a correlation and coupling between the electrical situations of upper and lower level distribution boxes, lacking the ability to predict load timing evolution, and lacking a hierarchical collaborative decision-making mechanism. Essentially, they lack systematic management technology for multi-level power distribution.

[0003] Most publicly available patents and academic research on similar power distribution control currently focus only on single-point monitoring and independent protection of a single distribution box, failing to construct a multi-level linkage situational awareness system based on a hierarchical power distribution network. At the communication level, most employ single communication modes such as power line carrier or LoRa, without considering channel attenuation characteristics and building interference to build dual-mode fault-tolerant coupled communication, resulting in deficiencies in channel stability and real-time transmission. At the architecture level, they generally adopt a highly coupled integrated design of acquisition, computation, and execution, without introducing hierarchical decoupling control. This leads to limited sensing accuracy, insufficient algorithm computing power, and delayed execution response, making it impossible to achieve upper-level pre-judgment control based on the evolution of lower-level electrical situations. It can only achieve post-fault handling, lacking active prevention and control support.

[0004] At the level of power system control algorithms, the core theoretical deficiencies of existing technologies can be summarized in three points: First, insufficient load characteristic identification. The third-level terminal distribution boxes cannot distinguish the differentiated electrical characteristics of lighting, air conditioning, and socket loads based on load impedance characteristics, harmonic spectrum characteristics, and time-series fluctuation patterns. They rely solely on a single current threshold to determine faults and cannot identify latent or gradual fault precursors. Second, a lack of hierarchical situational integration. The second- and first-level distribution boxes do not aggregate and integrate with the electrical data of multiple nodes at the lower level. They cannot analyze the superposition effect of multiple branch loads, fault propagation trends, and cumulative deterioration patterns of power quality. They can only passively receive fault reports, which can easily lead to overstepping of protection levels and large-scale power outages in the region. Third, a lack of dynamic predictive control. Traditional power distribution systems use a fixed threshold judgment mechanism and do not combine time-series evolution characteristics, hierarchical coupling correlation, and scene power consumption patterns to build adaptive predictions. This makes it impossible to achieve proactive management of potential hazards such as load peak impacts, harmonic accumulation, and latent leakage, resulting in bottlenecks in the safety, steady-state, and energy-saving aspects of the power distribution system.

[0005] Therefore, there is an urgent need for a collaborative control method that can realize hierarchical networking of three-level distribution boxes, dual-mode communication coupling, hierarchical situational awareness, and step-by-step prediction and pre-action. Based on the three-level architecture, a full-link intelligent power distribution management and control system can be realized, which includes end-level prediction, middle-level linkage, and top-level global control. Summary of the Invention

[0006] The purpose of this invention is to provide a building multi-level power distribution collaborative prediction system and control method with coupled networking, addressing the above-mentioned shortcomings, and solving the defects of existing building three-level power distribution systems such as hierarchical fragmentation, weak communication correlation, lagging protection mechanisms, and lack of hierarchical prediction support.

[0007] This invention is achieved through the following scheme: A coupled network building multi-level power distribution collaborative prediction system includes a sensing layer, an algorithm control layer, and an execution layer. The sensing layer is based on high-frequency synchronous acquisition of power parameters and fusion of multi-source sensors, and is deployed in each level of distribution box and each end load circuit within the distribution box; it integrates various detection units. The algorithm control layer is based on edge computing power scheduling theory, load feature deep learning model, hierarchical situation fusion algorithm, and time series trend prediction theory, and is deployed in independent embedded edge computing units in each level of distribution box; The execution layer consists of intelligent switches in each level of distribution box, graded circuit breakers, and load regulation modules, and only executes the actions issued by the algorithm control layer in a one-way manner.

[0008] The sensing layer includes various detection units such as PT voltage sensing unit, CT current sensing unit, FFT harmonic spectrum acquisition unit, temperature sensing unit, and dual-mode communication transmission unit.

[0009] The perception layer specifically includes the following models: Synchronous sampling model: Assume the time-domain voltage signal of the power distribution circuit is The time-domain current signal is An equal-interval synchronous sampling mechanism is adopted, and the sampling period is defined as follows: The number of sampling points per power frequency cycle is The discrete sampling sequence model is:

[0010] Parameter definition: The instantaneous voltage of the circuit is expressed in volts (V). The instantaneous current in the circuit is expressed in amperes (A). The sampling period is expressed in seconds (s). For discrete sampling sequence numbers; These are the discrete voltage and current sample values ​​at point k; This represents the total number of sampling points in a single cycle.

[0011] This solution also provides a method for coordinated predictive control of multi-level power distribution in coupled networks in buildings, including the following steps: S1. Initialize the global topology set of level 1 to 3 distribution boxes. Complete the PLC / LoRa dual-mode communication networking binding and link parameter matching, and build a multi-level power distribution hierarchical collaborative management and control topology architecture; S2 and the three-level distribution box rely on the steady-state reference 3σ model and introduce the anomaly deviation rate. With current harmonic distortion rate Dual core criteria accurately identify minute electrical anomalies at the end, enabling early prediction and on-site treatment of underlying hidden dangers, and preventing the upward transmission of anomalies; S3, Average status of tertiary distribution boxes under the aggregation of secondary distribution boxes Load fluctuation variance Combined with the time series rate of change Analyze the regional fault evolution and upward trend, and complete the mid-level gradient collaborative prevention and control and hierarchical fault blocking; S4, Level 1 distribution box relies on the total load of the entire area. Load fluctuation coefficient And voltage deviation quantitative indicators, comprehensively and quantitatively assess the steady-state risks of the power distribution system, and realize the pre-control of top-level systemic hidden dangers and peak load risks; S5, based on the dual-mode link signal-to-noise ratio threshold Intelligent switching criteria, based on uplink situation vector With downlink instruction vector This enables bidirectional synchronization of data and instructions between upper and lower levels, as well as seamless fault-tolerant switching of communication links. S6. Based on a time-series machine learning iterative model, dynamically update the steady-state mean. with standard deviation The core benchmark parameters are used to adaptively correct the system's prediction thresholds and control strategies, enabling long-term intelligent iterative optimization of power distribution management.

[0012] S1 specifically refers to: hierarchical architecture deployment and three-level power distribution topology network construction; Complete the hierarchical topology construction of the building's primary main distribution box, secondary distribution boxes, and tertiary terminal distribution boxes, and construct a hierarchical topology set quantitative model:

[0013] Parameter definition: A set of power distribution topologies covering the entire region; It is a collection of primary, secondary, and tertiary distribution boxes; For the corresponding level of distribution box equipment unit; The system automatically identifies hierarchical relationships and completes power line carrier channel matching and LoRa node networking binding, providing a topological foundation for subsequent hierarchical data synchronization and collaborative prediction.

[0014] Specifically, S2 is: As the end-point sensing and primary decision-making unit, the three-level distribution box is based on the differential load electrical characteristic identification and time-series micro-variation extraction algorithm. It constructs steady-state benchmark models, abnormal deviation models and graded fault criteria for three types of circuits: lighting, air conditioning and sockets, to achieve threshold-free pre-judgment. Among them, the load steady-state reference 3σ model;

[0015] Steady-state reference range:

[0016] Parameter definition: These are the steady-state sample parameters; Total sample size; It is the steady-state mean; The unbiased standard deviation; This is the normal operating range of the load. The time-series variation deviation rate (core predictive indicator) is:

[0017] Parameter definition: For real-time electrical parameters; The deviation rate is the rate of variation. This is a constant to prevent overflow.

[0018] The quantification criteria for the fourth-level state of the three-stage bin are as follows:

[0019] Parameter definition: The threshold for deviation rate classification; The threshold for harmonic classification; Based on the prediction of time-series evolution trends, the three-level box performs preemptive actions such as no-load energy consumption suppression, harmonic dynamic management, abnormal circuit current limiting, and local early warning of hidden dangers before the fault electrical quantity reaches the traditional protection threshold. This enables early identification, early intervention, and early suppression of micro-hidden dangers at the end, and blocks the transmission of abnormal electrical quantities from the bottom layer to the upper level from the source.

[0020] Specifically, S3 refers to: the secondary distribution box relies on the aggregation and fusion of multi-level power distribution status and the gradient evolution of fault propagation to gather data from the tertiary boxes under its jurisdiction. By aggregating the mean, fluctuation variance, and time series change rate, it quantitatively judges the abnormal upward trend in the region and realizes mid-level forward prevention and control. The three-level box cluster situation aggregation model is as follows:

[0021] Parameter definition: The number of containers under its jurisdiction at the third level; These are the parameters for a single tertiary box; This is the aggregated mean of the partitions; The variance of load fluctuation in the zone is denoted as .

[0022] The hierarchical fault propagation trend model is as follows:

[0023] Parameter definition: The rate of change over time; This refers to the duration of the sliding window. This refers to the current moment. The risk of hazard spreading upwards in the area was determined in a timely manner; When the algorithm determines that there is an abnormal upward trend, the accumulation of hidden dangers, and the deterioration of the situation in the lower-level tertiary power distribution area, the secondary distribution box will perform pre-control actions in advance, such as zonal load shifting, branch power threshold constraint, abnormal branch dynamic locking, and hierarchical reactive power compensation optimization. This will proactively block hierarchical faults caused by the superposition of abnormalities in multiple branches at the end, and avoid problems such as overload tripping of the secondary distribution box, overstepping of protection between upper and lower levels, and large-scale power outages in the zonal area.

[0024] Specifically, S4 refers to the primary distribution box, which serves as the core of the overall management and control of building power distribution. Based on the steady-state analysis of large-scale power distribution networks, the fusion of big data on the overall situation, and the prediction of systemic risks, it constructs quantitative risk criteria for load, fluctuation coefficient, and voltage deviation across the entire area, thereby achieving proactive management of systemic risks. The total load quantification model for the entire domain is as follows:

[0025] Parameter definition: Total active load for the entire building area; This represents the total number of secondary containers; For a single secondary box, the aggregation load is calculated. The model for the overall load fluctuation coefficient is as follows:

[0026] Parameter definition: This is the load fluctuation coefficient; This represents the maximum, minimum, and average load values ​​for the entire area within the window.

[0027] The criteria for classifying systemic risks across the entire domain are as follows:

[0028] Parameter definition: This is the load fluctuation threshold; This refers to voltage fluctuation. Rated voltage; Voltage offset grading threshold; Before the systemic risks of distribution steady-state imbalance, tiered fault accumulation, and power grid disturbance transmission take shape, the primary distribution box implements a top-level pre-control strategy that includes dynamic current limiting of the main incoming line, peak shaving and valley filling of the entire load, power constraints in high-risk zones, electrical isolation of tiered faults, and unified optimization of power quality across the entire network. This prevents problems such as cascading tripping of the main distribution cabinet, large-scale power outages in entire buildings, and distribution network instability.

[0029] Specifically, S5 involves defining bidirectional data vectors between upper and lower levels based on distributed multi-level system collaboration to achieve standardized transmission of situation and commands.

[0030] Parameter definition: Upload situational vectors to lower levels; To issue control vectors to the superior authority; This is the power limit value; This is the voltage reference value; This refers to the running status instructions; The system monitors the status of the dual-mode communication link in real time. When one link fails, the other link seamlessly takes over, completes the data, and resumes the transmission of instructions, ensuring the uninterrupted operation of the multi-level prediction logic and collaborative control mechanism.

[0031] Specifically, S6 involves using a rolling iterative algorithm based on time-series machine learning to dynamically correct steady-state baseline parameters, thus addressing the poor adaptability of traditional fixed threshold methods.

[0032]

[0033] Parameter definition: The average value before and after the iteration; The standard deviation before and after iteration; The forgetting factor controls the weight of historical data; The system continuously iterates and optimizes the anomaly detection thresholds, trend analysis weights, and pre-action strategies of each level of distribution box, so as to achieve long-term adaptive control of the three-level boxes adapting to the end load, the two-level boxes adapting to the zone load, and the one-level boxes adapting to the power consumption of the whole area.

[0034] In summary, due to the adoption of the above technical solution, the beneficial effects of this solution are: 1. This invention relies on a three-layer decoupled hierarchical architecture consisting of a perception layer, an algorithm control layer, and an execution layer. It constructs a dual-mode coupled channel transmission model based on wired power line carrier and LoRa wireless blind spot filling. It establishes a step-by-step prediction and quantification system of "end-end load feature identification - mid-level situation aggregation and judgment - top-level global steady-state optimization". This enables the three-level distribution boxes to take proactive actions based on the electrical evolution characteristics of the end load, the two-level distribution boxes to predict the situation gradient based on the three-level boxes' aggregation, and the first-level distribution boxes to globally control the situation based on the global hierarchical coupling. This changes the shallow control logic of traditional power distribution's "fault-based passive protection" and constructs a new intelligent quantitative management and control system for multi-level building power distribution networks that features proactive prediction, step-by-step prevention and control, global collaboration, and adaptive iteration. 2. This invention constructs a multi-level building power distribution hierarchical predictive collaborative control quantitative system, with a complete set of mathematical models for signal sampling, power calculation, channel transmission, situation fusion, timing prediction, and adaptive iteration. It upgrades power distribution management from "post-fault handling" to "pre-emptive hazard prediction and hierarchical collaborative governance", optimizing the quantitative control problem of proactive collaborative prevention and control of multi-level nested power distribution networks.

[0035] 3. A power line carrier + LoRa dual-mode coupled fault-tolerant communication quantization is proposed, and a mathematical model of channel attenuation, wireless propagation and intelligent switching is established. The communication characteristics under complex building obstruction environment are quantitatively analyzed to solve the problems of unstable communication and data asynchrony of multi-level power distribution nodes, and to ensure millisecond-level high-reliability operation of hierarchical prediction and collaborative control.

[0036] 4. Strictly implement layered decoupling control. All three layers of the architecture are equipped with dedicated quantitative calculation models to achieve decoupling of perception and data acquisition, algorithm analysis, and execution control, thereby improving perception accuracy, analysis accuracy, and command response speed.

[0037] 5. Establish a three-level quantitative judgment system: end-load micro-identification, mid-level zone situation fusion, and top-level global steady-state control. Through differentiated load models, cluster aggregation models, and global risk models, it can achieve step-by-step prediction and layer-by-layer defense, accurately block the transmission of fault levels, and completely solve the core pain points of traditional power distribution misjudgment, lag, and cascading tripping, thus significantly improving the stability and safety of building power distribution systems.

[0038] 6. By introducing a time-series machine learning adaptive iterative quantization model, the bottleneck of traditional fixed threshold control is broken. The judgment benchmark is updated in real time through dynamic iterative formula, and the system can adaptively adapt to the power consumption patterns of different scenarios, different time periods, and different loads. Attached Figure Description

[0039] Figure 1 This is a flowchart illustrating the multi-level power distribution fault prediction and blocking process of the present invention. Figure 2 This is a diagram of the overall three-layer decoupled architecture of the system of the present invention; Figure 3 for Figure 2 PLC+LoRa dual-mode coupled network topology diagram; Figure 4 This is a flowchart illustrating the multi-level power distribution fault prediction and blocking process of the present invention. Detailed Implementation

[0040] All features disclosed in this specification, or steps in all methods or processes disclosed herein, may be combined in any way, except for mutually exclusive features and / or steps.

[0041] Any feature disclosed in this specification (including any appended claims and abstract) may be replaced by other equivalent or similar features, unless specifically stated otherwise. That is, unless specifically stated otherwise, each feature is merely one example of a series of equivalent or similar features.

[0042] In the description of this invention, it should be understood that the terms "upper," "lower," "left," "right," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are only for the convenience of describing this invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a predetermined orientation, or be constructed and operated in a predetermined orientation. Therefore, they should not be construed as limitations on this invention.

[0043] Furthermore, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Therefore, a feature defined with "first," "second," etc., may explicitly or implicitly include one or more of that feature.

[0044] Example 1 like Figure 1 As shown, the present invention provides a technical solution: A coupled network building multi-level power distribution collaborative prediction system includes a sensing layer, an algorithm control layer and an execution layer. The sensing layer is based on high-frequency synchronous acquisition of power parameters and fusion of multi-source sensors. It is deployed in each level of distribution box and each end load circuit in the distribution box, and integrates PT voltage sensing unit, CT current sensing unit, FFT harmonic spectrum acquisition unit, temperature sensing unit and dual-mode communication transmission unit. The algorithm control layer, as the core decision-making center of the system, is based on edge computing power scheduling theory, load feature deep learning model, hierarchical situation fusion algorithm, and time series trend prediction theory, and is deployed in independent embedded edge computing units in each level of distribution box. The execution layer follows the principle of precise command execution and rapid action response. It consists of intelligent switches in distribution boxes at all levels, tiered circuit breakers, and load regulation modules. It only executes the current limiting, voltage regulation, load reduction, interlocking, tripping, and early warning actions issued by the algorithm control layer in a one-way manner. It does not participate in any data calculation, feature analysis, or logical decision-making. This achieves the architectural advantages of "decoupling decision-making and execution, and separating perception and control", ensuring that control commands are accurately implemented at the millisecond level and eliminating logical interference and action lag.

[0045] The perception layer specifically includes the following models; 1. Synchronous Sampling Model: Assume the time-domain voltage signal of the power distribution circuit is... The time-domain current signal is An equal-interval synchronous sampling mechanism is adopted, and the sampling period is defined as follows: The number of sampling points per power frequency cycle is The discrete sampling sequence model is:

[0046] Parameter definition: The instantaneous voltage of the circuit is expressed in volts (V). The instantaneous current in the circuit is expressed in amperes (A). The sampling period is expressed in seconds (s). For discrete sampling sequence numbers; These are the discrete voltage and current sample values ​​at point k; This represents the total number of sampling points in a single cycle.

[0047] 2. Quantitative Calculation Model for Basic Power Parameters: Periodic Average Active Power:

[0048] Periodic average reactive power:

[0049] Apparent power:

[0050] Power factor:

[0051] Parameter definition: Active power (W); Reactive power (Var); Apparent power (VA); Power factor; The current sampling value is lagging behind the power frequency cycle by 1 / 4.

[0052] 3. Harmonic distortion rate FFT quantization model: Fourier decomposition of electrical signals:

[0053] Total harmonic distortion of voltage:

[0054] Total harmonic distortion of current:

[0055] Parameter definition: It is the DC component of the voltage; The voltage and current amplitudes of the nth harmonic; It is the power frequency angular frequency; The initial phase of the nth harmonic; These are the fundamental voltage and current amplitudes. This refers to the harmonic distortion rate of voltage and current.

[0056] Example 2 like Figure 1 As shown, the present invention provides a technical solution: A method for coordinated predictive control of multi-level power distribution in a coupled network building, comprising the following steps: S1. Initialize the global topology set of level 1 to 3 distribution boxes. Complete the PLC / LoRa dual-mode communication networking binding and link parameter matching, and build a multi-level power distribution hierarchical collaborative management and control topology architecture; S2 and the three-level distribution box rely on the steady-state reference 3σ model and introduce the anomaly deviation rate. With current harmonic distortion rate Dual core criteria accurately identify minute electrical anomalies at the end, enabling early prediction and on-site treatment of underlying hidden dangers, and preventing the upward transmission of anomalies; S3, Average status of tertiary distribution boxes under the aggregation of secondary distribution boxes Load fluctuation variance Combined with the time series rate of change Analyze the regional fault evolution and upward trend, and complete the mid-level gradient collaborative prevention and control and hierarchical fault blocking; S4, Level 1 distribution box relies on the total load of the entire area. Load fluctuation coefficient And voltage deviation quantitative indicators, comprehensively and quantitatively assess the steady-state risks of the power distribution system, and realize the pre-control of top-level systemic hidden dangers and peak load risks; S5, based on the dual-mode link signal-to-noise ratio threshold Intelligent switching criteria, based on uplink situation vector With downlink instruction vector This enables bidirectional synchronization of data and instructions between upper and lower levels, as well as seamless fault-tolerant switching of communication links. S6. Based on a time-series machine learning iterative model, dynamically update the steady-state mean. with standard deviation Core benchmark parameters are used to adaptively correct the system's predicted thresholds and control strategies, enabling long-term intelligent iterative optimization of power distribution management and control. S1 specifically refers to: hierarchical architecture deployment and three-level power distribution topology network construction; Complete the hierarchical topology construction of the building's primary main distribution box, secondary distribution boxes, and tertiary terminal distribution boxes, and construct a hierarchical topology set quantitative model:

[0057] Parameter definition: A set of power distribution topologies covering the entire region; It is a collection of primary, secondary, and tertiary distribution boxes; For the corresponding level of distribution box equipment unit; The system automatically identifies hierarchical relationships and completes power line carrier channel matching and LoRa node networking binding, providing a topological foundation for subsequent hierarchical data synchronization and collaborative prediction.

[0058] Specifically, S2 refers to: pre-judgment of the terminal micro-state based on the load characteristic spectrum of the three-level distribution box (terminal quantitative prediction model). As the end-point sensing and primary decision-making unit, the three-level distribution box is based on the differential load electrical characteristic identification and time-series micro-variation extraction algorithm. It constructs steady-state benchmark models, abnormal deviation models and graded fault criteria for three types of circuits: lighting, air conditioning (including equipment) and sockets, to achieve threshold-free pre-judgment. Among them, the load steady-state reference 3σ model;

[0059] Steady-state reference range:

[0060] Parameter definition: These are the steady-state sample parameters; Total sample size; It is the steady-state mean; The unbiased standard deviation; This is the normal operating range of the load. The time-series variation deviation rate (core predictive indicator) is:

[0061] Parameter definition: For real-time electrical parameters; The deviation rate is the rate of variation. This is a constant to prevent overflow.

[0062] The quantification criteria for the fourth-level state of the three-stage box are as follows:

[0063] Parameter definition: The threshold for deviation rate classification; The threshold for harmonic classification; Based on the prediction of time-series evolution trends, the three-level box performs preemptive actions such as no-load energy consumption suppression, harmonic dynamic management, abnormal circuit current limiting, and local early warning of hidden dangers before the fault electrical quantity reaches the traditional protection threshold. This enables early identification, early intervention, and early suppression of micro-hidden dangers at the end, and blocks the transmission of abnormal electrical quantities from the bottom layer to the upper level from the source.

[0064] Specifically, S3 refers to: the secondary distribution box based on multi-node situational fusion-based mid-level gradient prediction and collaboration (partition aggregation prediction model): The secondary distribution box relies on the aggregation and fusion of multi-level power distribution status and the gradient evolution of fault propagation to gather data from the tertiary boxes under its jurisdiction. By aggregating the mean, fluctuation variance, and time series change rate, it quantitatively judges the abnormal upward trend in the region and realizes mid-level forward prevention and control.

[0065] The three-level box cluster situation aggregation model is as follows:

[0066] Parameter definition: The number of containers under its jurisdiction at the third level; These are the parameters for a single tertiary box; This is the aggregated mean of the partitions; The variance of load fluctuation in the zone.

[0067] The hierarchical fault propagation trend model is as follows:

[0068] Parameter definition: The rate of change over time; This refers to the duration of the sliding window. This refers to the current moment. The risk of hazard spreading upwards in the area was determined in a timely manner.

[0069] When the algorithm determines that there is an abnormal upward trend, the accumulation of hidden dangers, and the deterioration of the situation in the lower-level tertiary power distribution area, the secondary distribution box will perform pre-control actions such as zonal load shifting, branch power threshold constraint, abnormal branch dynamic locking, and hierarchical reactive power compensation optimization in advance. This will proactively block hierarchical faults caused by the superposition of abnormalities in multiple branches at the end, and avoid problems such as overload tripping of the secondary distribution box, overstepping of protection between upper and lower levels, and large-scale power outages in the zonal area.

[0070] Specifically, S4 refers to: Top-level steady-state prediction and control of the primary distribution box based on the global coupling situation (global steady-state risk model). As the core of building power distribution management, the primary distribution box, based on the steady-state analysis of large power distribution networks, the fusion of big data on the overall situation, and the prediction of systemic risks, constructs quantitative risk criteria for load, fluctuation coefficient, and voltage deviation across the entire area, thereby achieving proactive management of systemic risks. The total load quantification model for the entire domain is as follows:

[0071] Parameter definition: Total active load for the entire building area; This represents the total number of secondary containers; For a single secondary box, the aggregation load is calculated. The model for the overall load fluctuation coefficient is as follows:

[0072] Parameter definition: This is the load fluctuation coefficient; This represents the maximum, minimum, and average load values ​​for the entire area within the window.

[0073] The criteria for classifying systemic risks across the entire domain are as follows:

[0074] Parameter definition: This is the load fluctuation threshold; This refers to voltage fluctuation. Rated voltage; Voltage offset grading threshold; Before systemic risks such as overall power distribution imbalance, tiered fault accumulation, and power grid disturbance transmission take shape, the primary distribution box implements top-level pre-control strategies such as dynamic current limiting of the main incoming line, peak shaving and valley filling of the entire load, power constraints in high-risk zones, electrical isolation of tiered faults, and unified optimization of power quality across the entire network, to prevent problems such as cascading tripping of the main distribution cabinet, large-scale power outages of the entire building, and instability of the power distribution network.

[0075] S5 specifically refers to: a hierarchical situational two-way synchronization and dual-mode fault-tolerant collaborative quantification mechanism; Based on distributed multi-level system collaboration, bidirectional data vectors are defined between upper and lower levels to achieve standardized transmission of situation and commands:

[0076] Parameter definition: Upload situational vectors to lower levels; To issue control vectors to the superior authority; This is the power limit value; This is the voltage reference value; This refers to the running status instructions; The system monitors the status of the dual-mode communication link in real time. When one link fails, the other link seamlessly takes over, completes the data, and resumes the transmission of instructions, ensuring the uninterrupted operation of the multi-level prediction logic and collaborative control mechanism.

[0077] Specifically, S6 refers to a dynamic threshold adaptive iterative quantization model based on machine learning. Based on time-series machine learning, a rolling iterative algorithm is used to dynamically correct steady-state baseline parameters, overcoming the poor adaptability of traditional fixed threshold methods.

[0078]

[0079] Parameter definition: The average value before and after the iteration; The standard deviation before and after iteration; The forgetting factor controls the weight of historical data.

[0080] The system continuously iterates and optimizes the anomaly detection thresholds, trend analysis weights, and pre-action strategies of each level of distribution box, so as to achieve long-term adaptive control of the three-level boxes adapting to the end load, the two-level boxes adapting to the zone load, and the one-level boxes adapting to the power consumption of the whole area.

[0081] This solution is based on a building power distribution channel attenuation model, wireless radio frequency propagation, and a dual-mode link hot standby fault-tolerant mechanism. It constructs a coupled networking theoretical system with power line carrier (PLC) wired transmission as the primary method and LoRa wireless transparent transmission as a secondary method for blind spot compensation. Specifically, the power line carrier establishes a wired transmission channel based on existing power distribution lines, avoids power frequency interference based on power line carrier impedance matching, and has advantages such as low latency, high reliability, and resistance to electromagnetic interference. The LoRa wireless communication is based on spread spectrum communication theory, compensating for channel attenuation defects caused by building obstructions, and features real-time hot standby and seamless switching between the two links.

[0082] (1) Power line carrier PLC channel attenuation:

[0083] Parameter definition: Total transmission loss of PLC (dB); The inherent base loss of the line (dB); The node transmission distance (m); Carrier frequency (Hz); This is the frequency attenuation coefficient; Additional loss (dB) due to building shading.

[0084] (2) LoRa wireless propagation loss model;

[0085] Parameter definition: For wireless transmission loss (dB); Communication distance (km); The LoRa operating frequency (MHz).

[0086] (3) Criteria for intelligent switching of dual-mode links;

[0087] Parameter definition: For PLC link signal-to-noise ratio; For LoRa link signal-to-noise ratio; The signal-to-noise ratio is the threshold for normal link operation.

[0088] The core predictive logic at each level in this scheme is (a progressively forward-looking innovation quantification mechanism). 1. Three-level distribution box: Based on the power characteristics of the terminal circuits of lighting, air conditioning (equipment) and sockets, and relying on the micro-quantitative prediction of η anomaly deviation rate and THDI harmonic index, it realizes on-site and proactive management of terminal hidden dangers and blocks the upward flow of abnormalities from the bottom layer. 2. Secondary distribution box: Based on the aggregated parameters X2, fluctuation variance σ22, and time-series trend rate λ of all the tertiary boxes under its jurisdiction, the problem of hidden dangers spreading at the zonal level is solved, and the collaborative protection of the middle level is achieved; 3. Primary Distribution Box: Based on the total load Ptotal, fluctuation coefficient δ, and voltage deviation index, the overall situation is quantified to solve the systemic risks of the building's overall power distribution system, peak load risks, and hierarchical cumulative fault risks, and to achieve global steady-state pre-control.

[0089] In this plan Figure 1 This is a flowchart for multi-level power distribution fault prediction and blocking. This flowchart corresponds to the core innovative logic of this invention: hierarchical prediction, collaborative control, fault-tolerant communication, and adaptive iteration. The process progresses step by step from top to bottom, forming a complete closed-loop control system of "topology construction - end prediction - mid-level collaboration - global regulation - communication fault tolerance - iterative optimization". In this plan Figure 2The system has a three-layer decoupled architecture diagram. This solution adopts a layered decoupled architecture of perception layer, algorithm control layer and execution layer. Each layer has independent functions and is specifically adapted to computing power. The bottom layer is for end-point perception and prediction, the middle layer is for partitioned collaboration, and the top layer is for full-domain control and linkage. This completely solves the problems of lag and misjudgment caused by the coupling of traditional power distribution acquisition, calculation and execution. In this plan Figure 3 for Figure 2 PLC+LoRa dual-mode coupled network topology diagram; The system uses the power line carrier PLC wired link as the core transmission channel and the LoRa wireless link as the hot standby and blind spot filling channel to build a multi-level redundant network of primary master node, secondary relay node and tertiary end slave node to solve the problems of communication disconnection and data asynchrony caused by building obstruction and line interference, and ensure the reliability of hierarchical collaborative control. In this plan Figure 4 This solution presents a multi-level power distribution fault prediction and blocking mechanism diagram. It constructs a three-level progressive defense prediction and control mechanism. The three-level boxes address minor anomalies at the source, the two-level boxes intercept regional fault propagation, and the first-level boxes mitigate systemic risks across the entire area. This layer-by-layer blocking of faults prevents them from ascending and completely overturns the traditional protection mode of post-fault tripping in power distribution.

[0090] 1. Three-level distribution box: Based on the power characteristics of the lighting, air conditioning, and socket terminal circuits, relying on... Anomaly deviation rate Microscopic quantitative prediction of harmonic indicators enables on-site and proactive treatment of potential end-stage hazards, preventing the upward flow of abnormalities from the underlying layers. 2. Secondary distribution box: Based on the aggregated parameters of all tertiary distribution boxes under its jurisdiction. Fluctuation and variance Time series trend rate Quantitative analysis helps address the spread of hidden dangers across different zones and levels, enabling collaborative protection at the middle level. 3. Primary distribution box: based on the total load of the entire area. Volatility coefficient Voltage deviation index quantifies the overall situation, solves the systemic risks of the building's overall power distribution system, peak load risks, and tiered cumulative fault risks, and achieves global steady-state pre-control.

[0091] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A building multi-level power distribution collaborative prediction system with coupled networking, characterized in that, It includes a perception layer, an algorithm control layer, and an execution layer. The perception layer is based on high-frequency synchronous acquisition of power parameters and fusion of multi-source sensors, and is deployed in distribution boxes at all levels and in each end load circuit within the distribution boxes; it integrates various detection units. The algorithm control layer is based on edge computing power scheduling theory, load feature deep learning model, hierarchical situation fusion algorithm, and time series trend prediction theory, and is deployed in independent embedded edge computing units in each level of distribution box; The execution layer consists of intelligent switches in each level of distribution box, graded circuit breakers, and load regulation modules, and only executes the actions issued by the algorithm control layer in a one-way manner.

2. The building multi-level power distribution collaborative prediction system with coupled networking according to claim 1, characterized in that: The sensing layer includes various detection units such as PT voltage sensing unit, CT current sensing unit, FFT harmonic spectrum acquisition unit, temperature sensing unit, and dual-mode communication transmission unit.

3. The building multi-level power distribution collaborative prediction system with coupled networking according to claim 1, characterized in that: The perception layer specifically includes the following models: Synchronous sampling model: Assume the time-domain voltage signal of the power distribution circuit is The time-domain current signal is An equal-interval synchronous sampling mechanism is adopted, and the sampling period is defined as follows: The number of sampling points per power frequency cycle is The discrete sampling sequence model is: Parameter definition: The instantaneous voltage of the circuit is expressed in volts (V). The instantaneous current in the circuit is expressed in amperes (A). The sampling period is expressed in seconds (s). For discrete sampling sequence numbers; These are the discrete voltage and current sample values ​​at point k; This represents the total number of sampling points in a single cycle.

4. A method for coordinated predictive control of multi-level power distribution in coupled networks of buildings, characterized in that: Includes the following steps: S1. Initialize the global topology set of level 1 to 3 distribution boxes. Complete the PLC / LoRa dual-mode communication networking binding and link parameter matching, and build a multi-level power distribution hierarchical collaborative management and control topology architecture; S2 and the three-level distribution box rely on the steady-state reference 3σ model and introduce the anomaly deviation rate. With current harmonic distortion rate Dual core criteria accurately identify minute electrical anomalies at the end, enabling early prediction and on-site treatment of underlying hidden dangers, and preventing the upward transmission of anomalies; S3, Average status of tertiary distribution boxes under the aggregation of secondary distribution boxes Load fluctuation variance Combined with the time series rate of change Analyze the regional fault evolution and upward trend, and complete the mid-level gradient collaborative prevention and control and hierarchical fault blocking; S4, Level 1 distribution box relies on the total load of the entire area. Load fluctuation coefficient And voltage deviation quantitative indicators, comprehensively and quantitatively assess the steady-state risks of the power distribution system, and realize the pre-control of top-level systemic hidden dangers and peak load risks; S5, based on the dual-mode link signal-to-noise ratio threshold Intelligent switching criteria, based on uplink situation vector With downlink instruction vector This enables bidirectional synchronization of data and instructions between upper and lower levels, as well as seamless fault-tolerant switching of communication links. S6. Based on a time-series machine learning iterative model, dynamically update the steady-state mean. with standard deviation The core benchmark parameters are used to adaptively correct the system's prediction thresholds and control strategies, enabling long-term intelligent iterative optimization of power distribution management.

5. The building multi-level power distribution collaborative predictive control method with coupled networking according to claim 4, characterized in that: S1 specifically refers to: hierarchical architecture deployment and three-level power distribution topology network construction; Complete the hierarchical topology construction of the building's primary main distribution box, secondary distribution boxes, and tertiary terminal distribution boxes, and construct a hierarchical topology set quantitative model: Parameter definition: For the entire power distribution topology set; It is a collection of primary, secondary, and tertiary distribution boxes; For the corresponding level of distribution box equipment unit; The system automatically identifies hierarchical relationships and completes power line carrier channel matching and LoRa node networking binding, providing a topological foundation for subsequent hierarchical data synchronization and collaborative prediction.

6. The building multi-level power distribution collaborative predictive control method with coupled networking according to claim 4, characterized in that: Specifically, S2 is: As the end-point sensing and primary decision-making unit, the three-level distribution box is based on the differential load electrical characteristic identification and time-series micro-variation extraction algorithm. It constructs steady-state benchmark models, abnormal deviation models and graded fault criteria for three types of circuits: lighting, air conditioning and sockets, to achieve threshold-free pre-judgment. Among them, the load steady-state reference 3σ model; Steady-state reference range: Parameter definition: These are the steady-state sample parameters; Total sample size; It is the steady-state mean; The unbiased standard deviation; This is the normal operating range of the load. The time-series variation deviation rate (core predictive indicator) is: Parameter definition: For real-time electrical parameters; The deviation rate is the rate of variation. This is a constant to prevent overflow. The quantification criteria for the fourth-level state of the three-stage bin are as follows: Parameter definition: The threshold for deviation rate classification; The threshold for harmonic classification; Based on the prediction of time-series evolution trends, the three-level box performs preemptive actions such as no-load energy consumption suppression, harmonic dynamic management, abnormal circuit current limiting, and local early warning of hidden dangers before the fault electrical quantity reaches the traditional protection threshold. This enables early identification, early intervention, and early suppression of micro-hidden dangers at the end, and blocks the transmission of abnormal electrical quantities from the bottom layer to the upper level from the source.

7. The building multi-level power distribution collaborative predictive control method with coupled networking according to claim 4, characterized in that: Specifically, S3 refers to: the secondary distribution box relies on the aggregation and fusion of multi-level power distribution status and the gradient evolution of fault propagation to gather data from the tertiary boxes under its jurisdiction. By aggregating the mean, fluctuation variance, and time series change rate, it quantitatively judges the abnormal upward trend in the region and realizes mid-level forward prevention and control. The three-level box cluster situation aggregation model is as follows: Parameter definition: The number of containers under its jurisdiction at the third level; These are the parameters for a single tertiary box; This is the aggregated mean of the partitions; Variance of load fluctuation in different zones; The hierarchical fault propagation trend model is as follows: Parameter definition: The rate of change over time; This refers to the duration of the sliding window. For the current moment; when The risk of hazard spreading upwards in the area was determined in a timely manner; When the algorithm determines that there is an abnormal upward trend, the accumulation of hidden dangers, and the deterioration of the situation in the lower-level tertiary power distribution area, the secondary distribution box will perform pre-control actions in advance, such as zonal load shifting, branch power threshold constraint, abnormal branch dynamic locking, and hierarchical reactive power compensation optimization. This will proactively block hierarchical faults caused by the superposition of abnormalities in multiple branches at the end, and avoid problems such as overload tripping of the secondary distribution box, overstepping of protection between upper and lower levels, and large-scale power outages in the zonal area.

8. The building multi-level power distribution collaborative predictive control method with coupled networking according to claim 4, characterized in that: Specifically, S4 refers to the primary distribution box, which serves as the core of the overall management and control of building power distribution. Based on the steady-state analysis of large-scale power distribution networks, the fusion of big data on the overall situation, and the prediction of systemic risks, it constructs quantitative risk criteria for load, fluctuation coefficient, and voltage deviation across the entire area, thereby achieving proactive management of systemic risks. The total load quantification model for the entire domain is as follows: Parameter definition: Total active load for the entire building area; This represents the total number of secondary containers; For a single secondary box, the aggregation load is calculated. The model for the overall load fluctuation coefficient is as follows: Parameter definition: This is the load fluctuation coefficient; The maximum, minimum, and average load values ​​within the window; The criteria for classifying systemic risks across the entire domain are as follows: Parameter definition: This is the load fluctuation threshold; This refers to voltage fluctuation. Rated voltage; Voltage offset grading threshold; Before the systemic risks of distribution steady-state imbalance, tiered fault accumulation, and power grid disturbance transmission take shape, the primary distribution box implements a top-level pre-control strategy that includes dynamic current limiting of the main incoming line, peak shaving and valley filling of the entire load, power constraints in high-risk zones, electrical isolation of tiered faults, and unified optimization of power quality across the entire network. This prevents problems such as cascading tripping of the main distribution cabinet, large-scale power outages in entire buildings, and distribution network instability.

9. A building multi-level power distribution collaborative predictive control method with coupled networking as described in claim 4, characterized in that: Specifically, S5 involves defining bidirectional data vectors between upper and lower levels based on distributed multi-level system collaboration to achieve standardized transmission of situation and commands. Parameter definition: Upload situational vectors to lower levels; To issue control vectors to the superior authority; This is the power limit value; This is the voltage reference value; This refers to the running status instructions; The system monitors the status of the dual-mode communication link in real time. When one link fails, the other link seamlessly takes over, completes the data, and resumes the transmission of instructions, ensuring the uninterrupted operation of the multi-level prediction logic and collaborative control mechanism.

10. A building multi-level power distribution collaborative predictive control method with coupled networking according to claim 4, characterized in that: Specifically, S6 involves using a rolling iterative algorithm based on time-series machine learning to dynamically correct steady-state baseline parameters, thus addressing the poor adaptability of traditional fixed threshold methods. Parameter definition: The average value before and after the iteration; The standard deviation before and after iteration; The forgetting factor controls the weight of historical data; The system continuously iterates and optimizes the anomaly detection thresholds, trend analysis weights, and pre-action strategies of each level of distribution box, so as to achieve long-term adaptive control of the three-level boxes adapting to the end load, the two-level boxes adapting to the zone load, and the one-level boxes adapting to the power consumption of the whole area.

Images