Intelligent switch cabinet power distribution intelligent management and control system with adaptive load regulation
Patent Information
- Application Number
- CN202611069082.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-07-17
- Publication Date
- 2026-09-15
Smart Images
Figure CN122763397A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to the field of intelligent control technology for switchgear power distribution, specifically to an intelligent control system for switchgear power distribution with adaptive load regulation. Background Technology
[0002] With the continuous improvement of the intelligence level of power distribution systems, intelligent switchgear has been widely used in parallel power distribution networks. Single-cabinet-level adaptive load regulation technology has gradually matured, and each switchgear can independently perform load transfer and balancing regulation based on its local operating status. However, in the scenario of multiple switchgears operating in parallel, existing technologies generally lack a global coordination and control mechanism. Each switchgear triggers regulation actions independently based on its own load deviation, and adjacent switchgears cannot perceive each other's regulation intentions. When multiple switchgears detect load deviations at the same time and adjust in opposite directions, reverse regulation conflicts are easily formed, causing the load to be repeatedly transferred between switchgears, resulting in systemic load oscillations. Such oscillations not only reduce the efficiency of load balancing regulation and cause unnecessary switching operation losses, but may also cause bus voltage fluctuations, affecting the power supply stability and operational reliability of the power distribution system.
[0003] Therefore, there is an urgent need for an intelligent management and control solution with global coordination capabilities to effectively suppress load oscillations and reverse conflicts in multi-cabinet parallel scenarios.
[0004] To address the aforementioned shortcomings, a technical solution is provided. Summary of the Invention
[0005] The purpose of this invention is to solve the problems of load oscillation and reverse conflict caused by independent regulation of multiple parallel intelligent switchgear, and thus propose an intelligent control system for power distribution of intelligent switchgear with adaptive load regulation.
[0006] An intelligent switchgear power distribution control system with adaptive load regulation includes: Multi-cabinet operation status acquisition module: Using the built-in electrical quantity sensor group of each parallel intelligent switch cabinet as the main acquisition body, it collects multi-source data of a single cabinet and performs time sequence alignment and structured data storage to build a multi-cabinet operation status matrix; Load oscillation risk prediction module: retrieves the multi-cabinet operation status matrix, calculates the load rate deviation and change rate of each cabinet after smoothing and filtering, predicts the deviation value of the next adjustment cycle, counts the number of conflicting pairs of reverse adjustment intentions, and outputs the oscillation risk warning level. Adjustment priority token allocation module: Based on the oscillation risk warning level and the real-time load deviation of each cabinet, a weighted competition mechanism is used to allocate unique adjustment priority tokens among multiple cabinets, and the remaining cabinets enter a coordination waiting state; Load Coordination Decomposition and Distribution Module: Based on the global load rate optimization target, the token holding cabinet decomposes the global required load transfer amount into differentiated adjustment sub-tasks for each cabinet, generates a coordinated adjustment instruction set, and synchronously distributes it to each parallel cabinet to execute the load transfer operation; Adjustment closed-loop verification and recovery module: After each cabinet completes the adjustment, it reports the steady-state operation data, calculates the global equilibrium residual and compares it with the convergence threshold. If convergence is achieved, it broadcasts the recovery token to restore local adjustment. If convergence is not achieved, it adaptively corrects the adjustable capacity parameters and enters the next round of token competition.
[0007] Furthermore, the specific operation steps of the load oscillation risk prediction module include: The multi-cabinet operation status matrix is retrieved, and the active power sampling sequence within the continuous sliding window is extracted in groups according to the cabinet number and smoothed by using an exponential weighted moving average algorithm. Calculate the load rate deviation of a single cabinet based on the rated capacity and smoothed power of each cabinet, and construct a global load rate deviation vector. The first-order difference operation is performed on the load factor deviation vector to obtain the deviation change rate sequence, and the load deviation value at the end of the next adjustment cycle is predicted by the linear extrapolation method. If the measured load rate deviation value and the predicted load deviation value are in the same direction and are greater than the preset fixed deviation judgment threshold, it is determined that the cabinet has an active balancing adjustment need, the current cabinet is marked as the adjustment intention to activate the cabinet and the corresponding adjustment direction scalar is obtained. For any two activation cabinets on the same busbar, determine the conflict of reverse adjustment intentions and count the total number of conflict of reverse adjustment intentions. The total number of conflict pairs with opposing adjustment intentions is normalized to obtain a conflict density index. Based on the conflict density index, the oscillation risk warning level is divided into three levels: low, medium, and high.
[0008] Furthermore, the specific operation steps of the load oscillation risk prediction module also include: Retrieve historical adjustment event logs, capture all switching action records within the preset historical window, count the number of times other cabinets are triggered within a short period of time after a single cabinet adjustment, and calculate the inter-cabinet adjustment mutual trigger coefficient based on the time decay coefficient. Construct a mutual trigger coefficient matrix between cabinets and use the power iteration method to calculate the spectral radius. When the spectral radius exceeds a preset threshold, it is marked as a high-incidence state of historical oscillation. When the system is in a state of high historical oscillation, the original conflict density index is penalized and corrected based on the number of oscillation events occurring within the historical window and the increasing historical penalty weight. The risk level is then reclassified based on the corrected final conflict density index.
[0009] Furthermore, the specific operation steps of the adjustment permission token allocation module include: When the oscillation risk reaches the medium-to-high risk triggering condition, the global control layer issues a token competition invitation frame, and each parallel switch cabinet calculates the competition weight based on its current operating status and reports it. The competition weight is a weighted composite of three sub-weights: load deviation, urgency, and cooling penalty, weighted by preset coefficients. The global control layer selects the cabinet with the highest competitive weight and issues the priority token for this round of adjustment. The token carries a unique token serial number, token issuance timestamp, and token validity period. Once the token is issued, the global control layer immediately sends a regulation prohibition frame to all parallel switchgear that does not hold a token. After receiving the regulation prohibition frame, each switchgear suspends its local adaptive regulation logic process, freezes all pending local switching command queues, and enters a coordination waiting state.
[0010] Furthermore, the specific operation steps of the adjustment permission token allocation module also include: The global control layer starts an independent watchdog timer when the token is issued. If no adjustment completion confirmation frame is received within the delay tolerance period after the token expires, the current token serial number is forcibly declared invalid and the corresponding cabinet is marked as a suspected fault. For cabinets in a suspected fault state, the corresponding cooling penalty sub-weight is set to zero in subsequent token competition. At the same time, a suspected fault penalty coefficient is added to the current cabinet's total competition weight to reduce the probability of obtaining a token again. The marking is lifted after the local self-inspection is completed and a self-inspection pass message is reported. When all parallel cabinets are in a suspected fault state, the system automatically downgrades to a tokenless fault-tolerant regulation mode. Each cabinet is only allowed to perform local load shedding protection operations, and cross-cabinet load transfer switching operations are prohibited.
[0011] Furthermore, the specific operation steps of the load coordination and decomposition distribution module are as follows: The switchgear holding the adjustment priority token sends an adjustment amount decomposition request to the global control layer. The global control layer extracts the latest multi-cabinet operation status matrix and reads the real-time smoothed active power, rated capacity and adjustable capacity upper and lower limit parameters of each cabinet. The optimization objective is to minimize the mean square deviation of the global load rate distribution after adjustment, and the constraints are global power conservation constraints and boundary constraints of adjustment of each cabinet. An optimization objective function is constructed. The constrained quadratic programming solver is invoked to iteratively solve the optimal global adjustment vector, and each component corresponds to the differentiated load transfer amount of each cabinet. A distributed soft-start method is adopted, which divides the adjustment of each cabinet into several sub-steps evenly according to the preset single-step maximum transfer power limit, and sets a bus voltage stability waiting confirmation mechanism between adjacent sub-steps; The step-by-step operation sequences of each cabinet are integrated to generate a coordinated adjustment instruction set. The instruction set header carries a valid token serial number and is synchronously distributed through the priority communication channel. Each cabinet verifies that the serial number is consistent before entering the instruction execution queue. If they are inconsistent, the instruction set is discarded and an abnormal token serial number alarm is reported to the global control layer.
[0012] Furthermore, the specific operation steps of the closed-loop verification and recycling module are as follows: After each parallel switch cabinet completes all sub-step circuit breaker operations, it waits for the bus voltage of its cabinet to stabilize before collecting the steady-state operation data frame after adjustment, and reports it to the global control layer along with the token sequence number and the actual execution completion timestamp, in order to construct the adjustment effect evaluation vector and solve the global equilibrium residual. Compare the global equilibrium residual with the preset convergence threshold: If the current adjustment is determined to be converged, broadcast a token recovery notice to all cabinets, each cabinet will release the coordination waiting state and restore its local adjustment capability, and simultaneously clear the oscillation warning and record the complete adjustment event. If it is determined that the current adjustment has not converged, the estimated upper and lower limits of the adjustable capacity of each cabinet will be adjusted online using a proportional-integral correction strategy. After the parameters are corrected, the token automatically expires and re-enters the token competition cycle, executing a new round of token allocation and coordination process; At the same time, a fault accumulation counter is maintained for each switchgear. If the number of accumulated faults exceeds the preset limit, the corresponding switchgear will be temporarily removed and an alarm will be pushed. After continuous and stable convergence, it can be re-applied to join the coordination and adjustment participation list.
[0013] Compared with the prior art, the beneficial effects of the present invention are: This invention assesses the density of inter-cabinet regulation conflicts in real time through a load oscillation risk prediction mechanism, classifies risk levels based on historical coupling relationships, and triggers a tokenized serial regulation mode to avoid load oscillation caused by independent reverse regulation of multiple cabinets, thereby effectively improving the operational stability of parallel power distribution systems. This invention employs constrained quadratic programming for global regulation optimization decomposition, combined with a distributed soft-start step-by-step execution strategy, to achieve globally optimal load balancing under the premise of satisfying power conservation and single-cabinet capacity constraints, thereby reducing bus voltage fluctuations during regulation and improving load balancing accuracy and regulation smoothness. This invention sets up a closed-loop verification and adaptive parameter correction mechanism. By verifying the effect after adjustment, the adjustable capacity estimate is dynamically corrected. Combined with the continuous non-converged fault elimination strategy, it adaptively adapts to changes in field conditions, enhances the system's adjustment robustness, and reduces the cost of manual parameter maintenance. Attached Figure Description
[0014] Figure 1 This is the overall system block diagram of the present invention. Detailed Implementation
[0015] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.
[0016] Example: like Figure 1 As shown, the intelligent switchgear power distribution control system with adaptive load regulation includes a multi-cabinet operation status acquisition module, a load oscillation risk prediction module, a regulation authority token allocation module, a load coordination decomposition and distribution module, and a regulation closed-loop verification and recovery module.
[0017] Multi-cabinet operation status acquisition module: The main data acquisition component is an electrical quantity sensor group built into each parallel intelligent switchgear, consisting of a three-phase current sensor, a bus voltage transmitter, active and reactive power acquisition units, and auxiliary contacts for the open and closed positions of circuit breakers. Based on a set unified sampling period, the three-phase current, bus voltage, active power, reactive power, and circuit breaker status within each switchgear are collected synchronously. All sensor groups are equipped with a local nanometer-level hardware clock chip, and each frame of collected data is bound to a globally unique cabinet number and a hardware timestamp. Using the high-precision synchronous clock at the global control layer as a unified reference time standard, the original timestamps collected from each cabinet are used for... Introducing a fixed clock offset compensation amount for a single cabinet Execution timing alignment conversion formula: ,in, This represents the aligned standard unified timestamp; Select all cabinets to meet the requirements Falling in the same synchronous sampling window Within the collected data packets, abnormal jump data with timestamps deviating from the sampling window threshold are removed, and effective data grouping of multiple cabinets from the same source is completed; The grouped data after time-series alignment is standardized with a unified field structure. Each data tuple in the database contains the following fields: cabinet number, alignment timestamp, three-phase current sequence, bus voltage sequence, active power, reactive power, and circuit breaker status. The time-series database is used to store data in ascending order of timestamp, and a joint index of cabinet number and alignment timestamp is established. Define the total number of cabinets in the parallel power distribution system as follows: Using individual cabinets as the row dimension of the matrix and electrical operation characteristics as the column dimension, a matrix is constructed. A multi-cabinet operation status matrix, where each row of the matrix corresponds to the complete real-time operation characteristics of one parallel switchgear. Total number of columns for electrical characteristics; The sensor group continuously collects data based on a preset sampling period. After each round of time alignment and structured data entry, a new version of the real-time multi-cabinet operation status matrix is generated. When the total number of cabinets changes due to the commissioning or decommissioning of parallel switch cabinets or bus topology switching, the matrix row vectors are automatically added or deleted, and the matrix dimension parameters are updated synchronously.
[0018] Load oscillation risk prediction module: The multi-cabinet operation status matrix constructed by the multi-cabinet operation status acquisition module is retrieved, and the active power sampling sequence of each cabinet in the continuous sliding event window is extracted in groups according to the cabinet number. To address short-term spike interference from the load, an exponentially weighted moving average algorithm is used to smooth the original active power sequence. The smoothing formula is as follows: ,in, Indicates the first intelligent switch cabinet, the first Each synchronous sampling time The corresponding smoothed active power, This represents the smoothing weighted attenuation coefficient, with a default value of 0.15. Maintenance personnel can adjust it within the range of 0.05-0.30 according to the on-site load characteristics. No. intelligent switch cabinet, the first Each synchronous sampling time The original measured active power, Indicates the first Intelligent switch cabinet, last synchronous sampling time Smooth active power; Based on the rated capacity of each cabinet The deviation of the single cabinet load rate is calculated from the current smoothed power trend value. The calculation formula is: ,in, This represents the arithmetic mean of the smoothed active power of all parallel switchgear units. Traverse all parallel cabinets and construct the load rate deviation vector. ; Perform a first-order difference operation on the load factor deviation vector according to the synchronization timestamp to obtain the deviation change rate sequence; use a linear extrapolation method to predict the first... The predicted load deviation value of each cabinet at the end of the next adjustment cycle The calculation formula is: ,in, This indicates the preset adjustment period, set to 1 second; Represented as the first The cabinet at the current moment The average rate of change of deviation within the first three sections (600-millisecond window) is calculated using the following formula: ,in, Indicates the first Each cabinet is indexed by time. The corresponding instantaneous load rate deviation change rate at the cross section Indicates the section index. , indicating the current cross-section, This indicates the previous section (i.e., 200 milliseconds ago). , indicating the first two sections (i.e., 400 milliseconds ago); Preset fixed deviation judgment threshold Synchronously compare the measured load rate deviation value Deviation from predicted load : when and The positive and negative signs are in the same direction, and , When both conditions are met, the first... Each cabinet has a need for active balancing adjustment. This cabinet is marked as the adjustment intention activation cabinet, and its corresponding adjustment direction scalar... ,in, Represents a symbolic function; The value can be a discrete set. , This indicates that the current load rate of the cabinet is lower than the global average, and an overload operation needs to be performed. This indicates that the current load rate of the cabinet is higher than the global average, and a load reduction operation needs to be performed. This indicates that the current deviation of the cabinet is zero, and there is no need for adjustment. For any two activation cabinets on the same busbar and A cabinet pair is considered to constitute a conflict of interest in reverse adjustment if and only if the following conditions are met simultaneously: Adjust in opposite directions: ; The adjustment magnitudes are close: the ratio of the predicted adjustment magnitudes of the two cabinets falls within the preset symmetry range. ,in, This represents the preset symmetry factor, which is set between 1.2 and 2.0 based on field control experience, with a typical value of 1.5. Indicates the first The expected regulating power of each cabinet is the active power that the cabinet needs to transfer to achieve load balancing; the calculation formula is: The positive or negative value indicates the direction of adjustment. To accommodate the increased load required for this cabinet, This cabinet needs to be unloaded; Indicates the first The expected adjustable power of each cabinet, defined rules and completely consistent; The global control layer iterates through all pairs of activated cabinets within the current adjustment cycle, accumulating the total number of pairs with conflicting reverse adjustment intentions. ; The total number of reverse adjustment intent conflict pairs will be compared with the total number of currently activated cabinets. After normalization, the conflict density index is obtained: ,in, This indicates that there are no conflicts in activating the cabinet. This indicates that there is a reverse conflict between all activation cabinets; Based on the conflict density index, the oscillation risk warning level is divided into the following three levels: Low risk: The load distribution is relatively coordinated, and the independent adjustment logic can be executed normally without triggering the token allocation process. Medium risk: There is a certain degree of conflict in regulatory intent. The global control layer records early warning information and increases the frequency of subsequent cross-section monitoring. The allocation of regulatory authority tokens is only triggered when the conflict continues for more than three consecutive cross-sections. High risk: There is a serious conflict in adjustment intentions. Allowing multiple cabinets to independently execute adjustment actions would greatly increase the probability of load oscillations. At this level, the global control layer immediately issues a freeze command to all parallel switch cabinets to force a halt to independent adjustment logic. Upon receiving the command, each cabinet suspends all local adaptive adjustment processes, and the entire system switches to a tokenized serial adjustment mode. The tokenized serial adjustment mode refers to a serialized coordinated control state in which the global control layer issues adjustment priority tokens to a unique cabinet through competitive weights. The cabinet holding the token leads the generation of a globally coordinated and consistent adjustment instruction set and issues it to each cabinet for synchronous execution. Cabinets that do not hold tokens only respond to the globally issued instructions and do not trigger any local autonomous adjustment actions. For specific implementation methods, refer to the adjustment permission token allocation module and the load coordination decomposition and issuance module. Retrieve historical adjustment event logs from local storage, with a duration of [duration missing]. Record all switching actions within the historical window and calculate the inter-cabinet regulation mutual triggering coefficient: ,in, Indicates the first After each cabinet performs a load regulation and switching operation, within a preset short-term threshold... Inside, directly trigger the first The total number of times each cabinet initiated adjustment actions; Indicates the first Each cabinet is in the historical statistics window The total number of all adjustment actions performed within the period. Represents an exponential function. This represents the time decay coefficient, which is fixed at 0.02 and is used to control the rate at which the time interval weakens the coupling strength. The larger the value, the faster the decay occurs due to the time interval. Representing all cabinets in history Adjustment triggers cabinet In a linked event, the average time interval between two adjustment actions; Based on the mutual triggering coefficient The inter-cabinet trigger coefficient matrix, its diagonal elements Set to zero; calculate the spectral radius of the cross-trigger coefficient matrix using the power iteration method; When the spectral radius is greater than the preset threshold, it indicates that there is a strong historical coupling relationship between the adjustment actions of each cabinet, and the system is prone to oscillation under the condition of chain triggering. Therefore, the current state is marked as a high-incidence state of historical oscillation. When the system state is marked as a high-incidence historical oscillation state, a historical penalty weight is added to the conflict density index. The rule for determining this weight is: in the past Each recorded oscillation event occurs within the window. Increment by 0.1, with a maximum cap of 0.5; The criteria for determining an oscillation event are as follows: if the system enters a high-risk warning state three or more times within any consecutive 10-second time window, or if the active power fluctuation of any cabinet is actually monitored to exceed its rated capacity ±15% and the frequency is within the range of 0.5-5Hz for more than 2 seconds, it is determined to be an oscillation event. The conflict density index after penalty correction is: ; by Replace the original The risk warning level for oscillations will be reclassified; at the same time, the historical penalty weight will be adjusted accordingly. The window scrolls and updates, and when the system runs stably for more than 30 minutes without any new oscillation events, the penalty weight gradually decays to zero.
[0019] Adjusting the permission token allocation module: When the oscillation risk warning level reaches the medium-risk continuous trigger condition or the high-risk immediate trigger condition, the global control layer initiates the adjustment authority token allocation process: After receiving the token competition invitation frame issued by the global control layer, each parallel intelligent switch cabinet calculates the competition weight based on its current operating status. And report it; the competition weight is a weighted composite of three sub-weights, and the composite formula is: ,in, The load deviation sub-weight is defined as the first... Absolute value of current load rate deviation for each cabinet The result obtained by normalizing by maximum and minimum across all activated cabinets; The urgency sub-weight is defined as the first... The result obtained by normalizing the absolute value of the current deviation change rate of each cabinet to the maximum and minimum values among all activated cabinets; The weight of the cooling penalty sub-weight is defined as the first... The ratio of the time interval from when a cabinet last held the adjustment priority token to the current time to the preset maximum cooldown period, with the default value of the maximum cooldown period being 60 seconds. These are configurable weighting coefficients for the load deviation sub-weight, urgency sub-weight, and cooling penalty sub-weight, respectively, and they satisfy the constraints. The default configuration values are 0.4, 0.4, and 0.2, respectively, and can be adjusted online by maintenance personnel according to the on-site working conditions. After receiving the competition weight reports from all participating cabinets, the global control layer selects the cabinet with the highest competition weight and issues it a priority token for this round of adjustment. If two or more cabinets have the same highest competition weight, the cabinet with the lower number will receive the token first. The issued priority token carries three inherent attributes: a unique token serial number, a token issuance timestamp, and a token validity period. Based on the historical average completion time of adjustment actions Adaptive setting, the calculation formula is: The average completion time of the historical adjustment actions The initial empirical value is set at 2.5 seconds, and it is updated every time a closed-loop adjustment is completed after the system starts running. This is the sliding average of the completion times of the last 10 adjustment actions; After the token is issued, the global control layer immediately sends a regulation prohibition frame to all parallel switchgear that does not hold a token. The regulation prohibition frame is transmitted independently through a dedicated priority communication channel, which is physically isolated from the regular status data channel to ensure the real-time transmission of the prohibition frame. After receiving the regulation prohibition frame, each switchgear immediately suspends its local adaptive regulation logic process, freezes all pending local switching command queues, and enters a coordination waiting state. During the coordination waiting state, each switchgear continues to perform electrical quantity acquisition and status reporting tasks to maintain the real-time update of the operating status matrix, but does not respond to any locally triggered load regulation requests. In response to abnormal scenarios where token holders cannot reclaim their tokens properly due to communication interruptions or hardware malfunctions during the token's validity period, the global control layer starts an independent watchdog timer when the token is issued. If the global control layer does not receive an adjustment completion confirmation frame from the token holder within the delay tolerance period after the token's validity period expires, it will forcibly declare the current token serial number invalid and mark the holder as a suspected fault. The delay tolerance period is fixed at 500 milliseconds. For cabinets in a suspected fault state, their cooling penalty sub-weight is forcibly set to zero in subsequent token contention, and a suspected fault penalty coefficient is added to the cabinet's total contention weight. The default value is 0.3, which reduces its overall competitive weight to: ; Reduce the probability of the cabinet obtaining a token again until the cabinet completes the local self-inspection process and reports the self-inspection pass message to the global control layer. Then the suspected fault mark is removed and the competition weight is restored to the normal calculation rules. The specific implementation process of the local self-inspection procedure is as follows: The cabinet control unit first sends a known test message to the internal communication interface and verifies the consistency of the loopback message; Perform a continuity test on the circuit breaker's opening and closing control circuit, i.e., apply a 5V low voltage to check whether the circuit resistance is within the normal range of 0.5-2Ω; An internal standard reference source was injected into the voltage / current sampling channel to verify that the sampling deviation was less than ±0.5%; the internal standard reference source was 50% of its rated value. If all the above tests pass, a self-test pass message is generated; otherwise, a self-test failure message is generated with a failure code and reported to the global control layer. When all parallel cabinets are in a suspected fault state, the system automatically downgrades to a tokenless fault-tolerant regulation mode. In this mode, each cabinet is only allowed to perform local load shedding protection operations, and any cross-cabinet load transfer switching operations are prohibited to ensure the bottom line of safe power distribution operation.
[0020] Load Coordination and Decomposition Distribution Module: Upon receiving the regulation priority token, the switchgear sending a regulation decomposition request frame to the global control layer; the global control layer extracts the latest version of the multi-cabinet operating status matrix, reads the real-time smoothed active power, rated capacity, and upper and lower limits of adjustable capacity of each cabinet, and constructs a coordinated regulation decomposition optimization problem; the optimization objective is to minimize the mean square deviation of the global load rate distribution after regulation, and the constraints include global power conservation constraints and boundary constraints of regulation of each cabinet, which are fully expressed as: Optimize the objective function: ; Constraints: ; in, This represents a vector consisting of the expected adjustment power of all cabinets. This indicates the preset target load balancing rate, which defaults to the current global average load rate. Indicates the first The maximum load reduction (negative value) and maximum load increase that each cabinet can perform in the current state; This indicates that the object is subject to or satisfies a constraint. This represents a global power conservation constraint; The global control layer has a built-in constrained quadratic programming solver, which is constructed based on the mathematical model of the standard quadratic programming problem, and its optimization variable is the adjustment amount of each cabinet. The optimization objective is the aforementioned objective function, and the constraints are global power conservation constraints and boundary constraints for the adjustment of each cabinet; the solver loads the current global load distribution vector. The rated capacity and adjustable capacity constraints of each cabinet, as well as the target balanced load rate, are used as input parameters. A standard quadratic programming algorithm (such as the effective set method, interior point method, or sequential quadratic programming method) is called to iteratively solve the optimization objective function, and the optimal global adjustment vector is output. Each component within the vector corresponds to the differentiated load transfer amount that each cabinet needs to perform in this round. Positive values indicate that the cabinet needs to increase its load capacity, while negative values indicate that the cabinet needs to reduce its load capacity. When the global control layer maps the optimal global adjustment quantity to the operating sequence of the circuit breakers in each cabinet, it uses a distributed soft-start method to perform load transfer: [The text abruptly ends here, likely due to an incomplete sentence or a formatting error.] Adjustment amount of each cabinet According to the preset single-step maximum transfer power limit Divide the process into several sub-steps, each of which is: ,in, This indicates rounding up; the transfer amount for each step is... ; A stability waiting confirmation mechanism is set between adjacent sub-steps: After the sub-step is completed, the measured values of the bus voltage of each cabinet are continuously collected. When the voltage fluctuation amplitude of all parallel buses is lower than 0.5% of the rated voltage and the duration of stability is not less than the preset duration, the execution of the next sub-step is triggered. The step-by-step load transfer sub-steps, soft-start execution methods, and dynamic stability waiting criteria for each segment interval are integrated to generate a step-by-step operation sequence for the circuit breaker specific to each cabinet; all cabinet operation sequences are summarized and uniformly packaged into a coordination and adjustment instruction set for this round; the header field of the instruction set is fixed to carry the currently valid token sequence number as an authorization credential; The global control layer will synchronously distribute the coordination and adjustment instruction set to each parallel switch cabinet through the priority communication channel. After receiving the instruction set, each cabinet will verify whether the token sequence number in the instruction set header is consistent with the valid token sequence number currently cached in the cabinet. Only if the verification is successful can the instruction set be entered into the instruction execution queue. If the verification is inconsistent, the instruction set will be discarded and an abnormal token sequence number alarm will be reported to the global control layer to prevent expired instructions from being mistakenly executed after the token has been reclaimed due to communication delays.
[0021] Adjust the closed-loop verification and recycling module: After each parallel switchgear completes all sub-step circuit breaker operations according to the coordinated adjustment instruction set, it waits for the bus voltage stability criterion of its own cabinet to be met, and then collects the steady-state operation data frame after adjustment; the data frame content includes the three-phase current, bus voltage, active power, reactive power and the final state of the circuit breaker after adjustment, and attaches the token sequence number of this instruction set and the actual execution completion timestamp, and reports it to the global control layer; After receiving the steady-state operation data frames reported by all cabinets participating in this round of adjustment, the global control layer constructs an evaluation vector for the adjustment effect: , ,in, No. The absolute deviation between the current actual load rate and the target balanced load rate of each cabinet For the first The measured steady-state active power reported after each cabinet is adjusted; Calculate the global equilibrium residual: And compare it with the preset convergence threshold: If the global equilibrium residual is less than the preset convergence threshold, the current round of adjustment is considered converged. The global control layer broadcasts a token retrieval notification frame to all parallel switchgear. The notification frame carries the token sequence number of this round and the convergence confirmation identifier. After receiving the token retrieval notification frame, each switchgear releases its coordination waiting state and restores the operation capability of its local adaptive adjustment logic. The global control layer simultaneously clears the current oscillation risk warning state and writes the complete record of this round of adjustment event into the historical adjustment event log. The record includes: the token-holding switchgear number and the vector of the coordinated adjustment quantity decomposition result. The actual completion timestamps of each cabinet, the total number of sub-steps, and the final value of the global equilibrium residual for this round; If the global equilibrium residual is greater than or equal to the preset convergence threshold, the current adjustment is deemed unconverged; the adaptive correction process for adjustment parameters is initiated. Based on the evaluation vector of the adjustment effect The residual components of each cabinet and the coordinated adjustment amount in this round The deviation relationship of each component is used to correct the estimated upper and lower limits of the adjustable capacity of each cabinet online using a proportional-integral correction strategy. The specific calculation formula is as follows: Set the first... The actual adjustment amount of each cabinet in this round is: absolute deviation The revised adjustable capacity limit is then updated as follows: ,in, The first The adjustable capacity limit of each cabinet before and after the update; This represents the proportionality coefficient, which is a fixed value of 0.5. This represents the integral coefficient, which is taken as a fixed value of 0.1. This indicates the number of consecutive non-converging cycles of the cabinet. Indicates the time interval between each adjustment round; Indicates the first The absolute deviation of the current cabinet; The revised adjustable capacity lower limit is updated as follows: ; Revised and Must meet , and Reasonable constraints; After the parameters are corrected, the tokens automatically become invalid and re-enter the token competition cycle. Each counter will re-report the competition weight based on the corrected status and execute a new round of token allocation and coordination adjustment process. The global control layer maintains a fault accumulation counter for each parallel switchgear, using the cabinet number as the key and the count of consecutive non-convergence cycles as the value. Each time the adjustment converges, all cabinet counters involved in this round of adjustment are reset to zero; Each time the adjustment fails to converge, the counters of all cabinets participating in this round of adjustment are incremented by 1; When the counter for consecutive non-convergence rounds of any cabinet exceeds the preset upper limit of 3 rounds, the global control layer temporarily removes the cabinet from the current coordination and adjustment participation list. The remaining cabinets are re-executed for coordination and adjustment quantity decomposition and optimization to eliminate the situation where the actual adjustable capacity of a single cabinet deviates significantly from the configuration parameters, thus continuously dragging down the global convergence and ensuring the overall system adjustment robustness. At the same time, the global control layer pushes an alarm for the adjustable capacity of the cabinet to the operation and maintenance terminal, prompting manual intervention to verify the consistency between the actual circuit switching status of the cabinet and the rated parameter configuration. After more than 5 consecutive stable convergence rounds, the cabinets that were temporarily removed can reapply to be added to the coordination and adjustment participation list, and will be restored by the global control layer at the start of the next token competition cycle.
[0022] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.
Claims
1. An intelligent switch cabinet power distribution intelligent management and control system with adaptive load regulation, characterized in that, include: Multi-cabinet operation status acquisition module: Using the built-in electrical quantity sensor group of each parallel intelligent switch cabinet as the main acquisition body, it collects multi-source data of a single cabinet and performs time sequence alignment and structured data storage to build a multi-cabinet operation status matrix; Load oscillation risk prediction module: retrieves the multi-cabinet operation status matrix, calculates the load rate deviation and change rate of each cabinet after smoothing and filtering, predicts the deviation value of the next adjustment cycle, counts the number of conflicting pairs of reverse adjustment intentions, and outputs the oscillation risk warning level. Adjustment priority token allocation module: Based on the oscillation risk warning level and the real-time load deviation of each cabinet, a weighted competition mechanism is used to allocate unique adjustment priority tokens among multiple cabinets, and the remaining cabinets enter a coordination waiting state; Load Coordination Decomposition and Distribution Module: Based on the global load rate optimization target, the token holding cabinet decomposes the global required load transfer amount into differentiated adjustment sub-tasks for each cabinet, generates a coordinated adjustment instruction set, and synchronously distributes it to each parallel cabinet to execute the load transfer operation; Adjustment closed-loop verification and recovery module: After each cabinet completes the adjustment, it reports the steady-state operation data, calculates the global equilibrium residual and compares it with the convergence threshold. If convergence is achieved, it broadcasts the recovery token to restore local adjustment. If convergence is not achieved, it adaptively corrects the adjustable capacity parameters and enters the next round of token competition.
2. The adaptive load-regulated intelligent switchgear power distribution intelligent management and control system according to claim 1, characterized in that, The specific operation steps of the load oscillation risk prediction module include: The multi-cabinet operation status matrix is retrieved, and the active power sampling sequence within the continuous sliding window is extracted in groups according to the cabinet number and smoothed by using an exponential weighted moving average algorithm. Calculate the load rate deviation of a single cabinet based on the rated capacity and smoothed power of each cabinet, and construct a global load rate deviation vector. The first-order difference operation is performed on the load factor deviation vector to obtain the deviation change rate sequence, and the load deviation value at the end of the next adjustment cycle is predicted by the linear extrapolation method. If the measured load rate deviation value and the predicted load deviation value are in the same direction and are greater than the preset fixed deviation judgment threshold, it is determined that the cabinet has an active balancing adjustment need, the current cabinet is marked as the adjustment intention to activate the cabinet and the corresponding adjustment direction scalar is obtained. For any two activation cabinets on the same busbar, determine the conflict of reverse adjustment intentions and count the total number of conflict of reverse adjustment intentions. The total number of conflict pairs with opposing adjustment intentions is normalized to obtain a conflict density index. Based on the conflict density index, the oscillation risk warning level is divided into three levels: low, medium, and high.
3. The intelligent switchgear power distribution intelligent control system with adaptive load regulation according to claim 2, characterized in that, The specific operation steps of the load oscillation risk prediction module also include: Retrieve historical adjustment event logs, capture all switching action records within the preset historical window, count the number of times other cabinets are triggered within a short period of time after a single cabinet adjustment, and calculate the inter-cabinet adjustment mutual trigger coefficient based on the time decay coefficient. Construct a mutual trigger coefficient matrix between cabinets and use the power iteration method to calculate the spectral radius. When the spectral radius exceeds a preset threshold, it is marked as a high-incidence state of historical oscillation. When the system is in a state of high historical oscillation, the original conflict density index is penalized and corrected based on the number of oscillation events occurring within the historical window and the increasing historical penalty weight. The risk level is then reclassified based on the corrected final conflict density index.
4. The intelligent switchgear power distribution intelligent control system with adaptive load regulation according to claim 1, characterized in that, The specific operation steps of the adjustment permission token allocation module include: When the oscillation risk reaches the medium-to-high risk triggering condition, the global control layer issues a token competition invitation frame, and each parallel switch cabinet calculates the competition weight based on its current operating status and reports it. The competition weight is a weighted composite of three sub-weights: load deviation, urgency, and cooling penalty, weighted by preset coefficients. The global control layer selects the cabinet with the highest competitive weight and issues the priority token for this round of adjustment. The token carries a unique token serial number, token issuance timestamp, and token validity period. Once the token is issued, the global control layer immediately sends a regulation prohibition frame to all parallel switchgear that does not hold a token. After receiving the regulation prohibition frame, each switchgear suspends its local adaptive regulation logic process, freezes all pending local switching command queues, and enters a coordination waiting state.
5. The intelligent switchgear power distribution control system with adaptive load regulation according to claim 4, characterized in that, The specific operation steps of the adjustment permission token allocation module also include: The global control layer starts an independent watchdog timer when the token is issued. If no adjustment completion confirmation frame is received within the delay tolerance period after the token expires, the current token serial number is forcibly declared invalid and the corresponding cabinet is marked as a suspected fault. For cabinets in a suspected fault state, the corresponding cooling penalty sub-weight is set to zero in subsequent token competition. At the same time, a suspected fault penalty coefficient is added to the current cabinet's total competition weight to reduce the probability of obtaining a token again. The marking is lifted after the local self-inspection is completed and a self-inspection pass message is reported. When all parallel cabinets are in a suspected fault state, the system automatically downgrades to a tokenless fault-tolerant regulation mode. Each cabinet is only allowed to perform local load shedding protection operations, and cross-cabinet load transfer switching operations are prohibited.
6. The intelligent switchgear power distribution control system with adaptive load regulation according to claim 1, characterized in that, The specific operation steps of the load coordination and decomposition distribution module are as follows: The switchgear holding the adjustment priority token sends an adjustment amount decomposition request to the global control layer. The global control layer extracts the latest multi-cabinet operation status matrix and reads the real-time smoothed active power, rated capacity and adjustable capacity upper and lower limit parameters of each cabinet. The optimization objective is to minimize the mean square deviation of the global load rate distribution after adjustment, and the constraints are global power conservation constraints and boundary constraints of adjustment of each cabinet. An optimization objective function is constructed. The constrained quadratic programming solver is invoked to iteratively solve the optimal global adjustment vector, and each component corresponds to the differentiated load transfer amount of each cabinet. A distributed soft-start method is adopted, which divides the adjustment of each cabinet into several sub-steps evenly according to the preset single-step maximum transfer power limit, and sets a bus voltage stability waiting confirmation mechanism between adjacent sub-steps; The step-by-step operation sequences of each cabinet are integrated to generate a coordinated adjustment instruction set. The instruction set header carries a valid token serial number and is synchronously distributed through the priority communication channel. Each cabinet verifies that the serial number is consistent before entering the instruction execution queue. If they are inconsistent, the instruction set is discarded and an abnormal token serial number alarm is reported to the global control layer.
7. The intelligent switchgear power distribution control system with adaptive load regulation according to claim 1, characterized in that, The specific operation steps of the closed-loop verification and recovery module are as follows: After each parallel switch cabinet completes all sub-step circuit breaker operations, it waits for the bus voltage of its cabinet to stabilize before collecting the steady-state operation data frame after adjustment, and reports it to the global control layer along with the token sequence number and the actual execution completion timestamp, in order to construct the adjustment effect evaluation vector and solve the global equilibrium residual. Compare the global equilibrium residual with the preset convergence threshold: If the current adjustment is determined to be converged, broadcast a token recovery notice to all cabinets, each cabinet will release the coordination waiting state and restore its local adjustment capability, and simultaneously clear the oscillation warning and record the complete adjustment event. If it is determined that the current adjustment has not converged, the estimated upper and lower limits of the adjustable capacity of each cabinet will be adjusted online using a proportional-integral correction strategy. After the parameters are corrected, the token automatically expires and re-enters the token competition cycle, executing a new round of token allocation and coordination process; At the same time, a fault accumulation counter is maintained for each switchgear. If the number of accumulated faults exceeds the preset limit, the corresponding switchgear will be temporarily removed and an alarm will be pushed. After continuous and stable convergence, it can be re-applied to join the coordination and adjustment participation list.