A kind of special variable acquisition terminal with intelligent regulation and control function of electric load
Patent Information
- Application Number
- CN202611222681.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2026-08-13
- Publication Date
- 2026-09-15
Smart Images
Figure SMS_38 
Figure SMS_48 
Figure SMS_58
Abstract
Description
Technical Field
[0001] This invention relates to the field of power consumption control technology. More specifically, this invention relates to a dedicated transformer data acquisition terminal with intelligent power load control function. Background Technology
[0002] Dedicated transformer data acquisition terminals are specialized devices for collecting electricity consumption information from dedicated transformer users. With the deepening of intelligent distribution network construction, these terminals are gradually extending beyond traditional functions such as power data acquisition and current / voltage monitoring to include intelligent dispatching. To achieve demand response, maximum demand control, and safe management of the distribution network for dedicated transformer users, intelligent load regulation has become an indispensable core function of dedicated transformer data acquisition terminals. This function is of great practical significance for optimizing power grid resource allocation and ensuring the safe operation of local distribution systems.
[0003] In existing load control practices, dedicated transformer acquisition terminals typically use control strategies that set fixed power thresholds or specific time period rules to limit users' electricity load. However, this traditional control method has obvious limitations. It can only simply determine whether the total load value exceeds the set limit or whether the growth rate is too fast, but it cannot reflect and identify the underlying physical reasons for load changes. Since most dedicated transformer users are large factories with complex power systems, in actual production processes, the load increase of normal production lines in workshops and the surge in transient current when heavy motors are powered on and started can all cause the overall power to fluctuate significantly in a short period of time.
[0004] Since existing technology cannot analyze the true causes of load fluctuations, traditional control strategies are prone to misjudging normal production operations as abnormal power consumption, which frequently triggers the underlying circuit breaker to erroneously control load limits and power outages. This not only seriously disrupts the normal production order of the factory and causes process interruptions, but also makes the final control results seriously contradict the actual power demand of users. Summary of the Invention
[0005] To address the technical problem that existing technologies fail to reflect the true causes of power load fluctuations, easily misinterpreting normal equipment operation as power anomalies, leading to false triggering of load limiting control and impacting normal production, this invention provides a dedicated transformer data acquisition terminal with intelligent power load regulation function. The terminal includes a memory, a processor, and a computer program stored in the memory and running on the processor. When the processor executes the computer program, it performs the following steps:
[0006] Acquire basic electrical data for the current time period and historical reference time periods. The basic electrical data includes three-phase voltage data, three-phase current data, three-phase active power data, and three-phase reactive power data. Determine the probability of total power anomaly and the probability of power anomaly in each phase based on the basic electrical data for the current time period, thereby determining whether the current time period is a suspected load anomaly period.
[0007] In response to the current period being a suspected period of abnormal load, valid event points are extracted based on basic electrical data, and the current period is divided into multiple event segments using the valid event points in the current period. The last event segment obtained from the current period is taken as the current event segment.
[0008] The historical reference period is spliced together with the period before the current event segment in the current period on the time axis to form a continuous long historical sequence interval, and the long historical sequence interval is divided into multiple historical event segments using effective event points;
[0009] Search for suspected similar historical windows that match the current event segment within a long historical sequence interval, and calculate the probability that the device corresponding to the current event segment is operating normally;
[0010] Based on the proportion of suspected similar historical windows contained in each historical event segment, identify historical event segments similar to the current event segment, and assess the overall importance of the equipment corresponding to the current event segment based on the distribution characteristics of similar historical event segments and the magnitude of changes in electrical data of the current event segment.
[0011] A load adjustment index is generated by combining the overall importance of the equipment and the probability of normal operation. The maximum allowable total power and the maximum limited power of each phase are adjusted based on the load adjustment index and then sent to the underlying intelligent equipment to perform power load regulation.
[0012] Preferably, the step of determining the probability of total power anomaly and the probability of power anomaly in each phase based on the basic electrical data of the current time period, and thereby determining whether the current time period is a suspected load anomaly period, includes:
[0013] For any monitoring sampling moment in the current time period, the ratio between the sum of the three-phase active power at that monitoring sampling moment and the maximum allowable total power set by the current system is taken as the total power risk level at that monitoring sampling moment; the total power risk levels of adjacent monitoring sampling moments in the current time period are subjected to discrete difference operation and normalization processing to obtain the total power trend risk level of the current time period; the average of the total power risk level of the last monitoring sampling moment in the current time period and the total power trend risk level of the current time period is taken as the total power anomaly probability of the current time period.
[0014] Extract the cluster centers of the three-phase power distribution ratio within the historical reference period as the normal benchmark ratio of three-phase power, and combine them with the maximum allowable total power set by the current system to obtain the maximum limited power of each phase.
[0015] For any phase, the ratio between the active power of the phase and the maximum limiting power of the phase at each monitoring sampling time within the current time period is taken as the power risk level of the phase at each monitoring sampling time within the current time period; discrete difference calculation and normalization are performed on the power risk levels of the phase at adjacent monitoring sampling times within the current time period to obtain the power trend risk level of the phase; the average value between the power risk level of the phase at the last monitoring sampling time of the current time period and the power trend risk level of the phase in the current time period is taken as the probability of power anomaly in the current time period.
[0016] The maximum value among the total power anomaly probability and the power anomaly probability of each phase is taken as the load anomaly probability; in response to the load anomaly probability being greater than or equal to the anomaly warning threshold, the current period is determined to be a suspected load anomaly period.
[0017] Preferably, the risk level of the total power trend satisfies the expression:
[0018] ;
[0019] in, This indicates the degree of risk associated with the total power trend during the current period; This indicates the total number of monitoring and sampling times included in the current time period; Indicates the first time in the current time period The overall power risk level at each monitoring sampling time; Indicates the first time in the current time period The overall power risk level at each monitoring sampling time; This represents the normalization function.
[0020] Preferably, the extraction of valid event points based on basic electrical data includes:
[0021] For each type of basic electrical data, the arithmetic mean of the absolute values of the changes in basic electrical data within the previous noise baseline window at each monitoring sampling time is extracted as the anti-interference baseline for dynamic fluctuations. Based on the difference between the changes in basic electrical data at each monitoring sampling time and the anti-interference baseline, the probability of each monitoring sampling time being an event point is obtained through multi-dimensional feature extreme value screening to form an event point probability sequence. The event point probability sequence is segmented using the maximum inter-class variance threshold segmentation method to extract the optimal truncation threshold. Monitoring sampling times with a probability greater than or equal to the optimal truncation threshold are determined as valid event points. In response to the existence of multiple adjacent valid event points on the time axis, only the monitoring sampling time with the earliest time series is retained as a valid event point.
[0022] Preferably, the probability that each monitoring sampling time is an event point satisfies the expression:
[0023] ;
[0024] in, Indicates the first time in the current time period The probability that a monitoring sampling time is an event point; Indicates the first time in the current time period The first monitoring sampling time Changes in basic electrical data; Indicates the total number of basic electrical data types; This indicates the extraction of the maximum value result from all types of basic electrical data; Indicates the absolute value symbol; Indicates the first Within the noise baseline window preceding the monitoring sampling time, the first The arithmetic mean of the absolute values of the changes in basic electrical data; This represents the constant for preventing overflow.
[0025] Preferably, the step of finding a potentially similar historical window that matches the current event segment within a long historical sequence interval, and calculating the probability that the device corresponding to the current event segment is operating normally, includes:
[0026] The basic electrical data type corresponding to the maximum value of the change in various basic electrical data at the first monitoring sampling time in the current event segment is taken as the target monitoring data. A historical sliding window with the same length as the current event segment is established and traversed in the long historical sequence interval. The similarity of the change trend of the target monitoring data between each historical sliding window and the current event segment is calculated, and a clustering algorithm is used to extract suspected similar historical windows. Based on the degree of change of each basic electrical data at the beginning of the current event segment, the influence weight of each basic electrical data on the normal operation of the equipment in the current event segment is determined. Combining the similarity of the change trend of each basic electrical data between each suspected similar historical window and the current event segment and the influence weight, the probability of the normal operation of the corresponding equipment in the current event segment is calculated.
[0027] Preferably, the probability that the device corresponding to the current event segment is operating normally satisfies the expression:
[0028] ;
[0029] in, This indicates the probability that the device corresponding to the current event segment is operating normally; This indicates the total number of potentially similar historical windows in the current event segment; Indicates the total number of basic electrical data types; Indicates the first The first suspected similar historical window and the current event segment Similarity in the changing trends of basic electrical data; Indicates the first The weight of the impact of basic electrical data on the normal operation of equipment in the current event segment.
[0030] Preferably, the step of identifying historical event segments similar to the current event segment based on the proportion of suspected similar historical windows contained in each historical event segment, and assessing the overall importance of the equipment corresponding to the current event segment based on the distribution characteristics of similar historical event segments and the magnitude of changes in the electrical data of the current event segment, includes:
[0031] The proportion of the number of suspected similar historical windows contained in each historical event segment within a long historical sequence interval to the total number of windows contained in that historical event segment is used as the probability that each historical event segment is similar to the current event segment. In response to the probability of similarity being greater than a set threshold, the corresponding historical event segment is determined to be similar to the current event segment, and consecutively adjacent similar historical event segments on the timeline are merged into complete similar event segments, while isolated similar historical event segments are directly recorded as complete similar event segments.
[0032] Based on the total number of complete similar event segments and the average duration of complete similar event segments within the entire historical long sequence interval, the temporal importance of the equipment corresponding to the current event segment is determined; the sum of the influence weights of various basic electrical data on the normal operation of the equipment in the current event segment is calculated and normalized to obtain the system influence importance of the equipment corresponding to the current event segment; the mean between the temporal importance and the system influence importance is taken as the comprehensive importance of the equipment corresponding to the current event segment.
[0033] Preferably, the step of generating a load adjustment index by combining the overall importance of the equipment with the probability of normal operation includes:
[0034] The overall importance of the equipment is normalized to the range of [-1, 1] to obtain the load adjustment intention parameter; in response to the load adjustment intention parameter being negative, the product of the difference between 1 and the probability of the equipment operating normally in the current event segment and the load adjustment intention parameter is used as the load adjustment index; in response to the load adjustment intention parameter being positive, the product of the probability of the equipment operating normally in the current event segment and the load adjustment intention parameter is used as the load adjustment index.
[0035] Preferably, the adjustment of the maximum allowable total power and the maximum limited power of each phase based on the load adjustment index includes:
[0036] The product of the current system's maximum permissible total power and the load adjustment index is taken as the maximum total power adjustment level. The maximum safe carrying capacity specified on the nameplate of the distribution transformer and the minimum required power of the factory's basic equipment are combined to apply boundary restrictions to the maximum total power adjustment level in order to obtain the adjusted maximum permissible total power.
[0037] Adjust the power limit for each phase separately based on the probability of power anomalies in each phase:
[0038] ;
[0039] in, This indicates the maximum power limit after adjustment for phase A; This indicates the maximum power limit for phase A; Indicates the degree of adjustment of the maximum total power; This indicates the probability of an abnormal power output in phase A during the current time period; This indicates the probability of abnormal power in phase B during the current time period; This indicates the probability of an abnormal C-phase power during the current time period.
[0040] The beneficial effects of this invention are as follows: When faced with a suspected load anomaly period, this invention assesses the probability of normal operation of the equipment causing the fluctuation and its overall importance by comparing the changes in the basic electrical data of the current event segment with similar historical windows in the long historical sequence interval. Based on this, a load adjustment index is dynamically generated to correct the maximum allowable total power and the maximum limited power of each phase. This allows for in-depth analysis of the true cause of the load surge through the data surface, distinguishing between the normal start-up of core equipment and the actual abnormal overload. Thus, while strictly adhering to the capacity limits of the distribution network, it provides differentiated available power expansion for important and normally operating equipment, and strictly limits current for non-core or suspected faulty equipment. This overcomes the shortcomings of traditional fixed threshold strategies, which are prone to misjudging normal production actions as abnormal due to the inability to identify the cause of load changes. It effectively avoids the false triggering of power outage tripping control by the dedicated transformer acquisition terminal, and greatly ensures the continuity and stability of the user's core production line while realizing intelligent power control. Detailed Implementation
[0041] The technical solutions in the embodiments of the present invention will be clearly and completely described below. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. 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.
[0042] The specific embodiments of the present invention are described in detail below.
[0043] This invention provides a dedicated transformer data acquisition terminal with intelligent power load control function, including a processor and a memory. The processor executes a computer program stored in the memory to implement an intelligent power load control method for the dedicated transformer data acquisition terminal. The method includes the following steps S1-S6:
[0044] S1: Obtain basic electrical data for the current time period and historical reference time periods.
[0045] Specifically, the voltage and current transformers integrated within the dedicated transformer data acquisition terminal are used to collect three-phase voltage and current data from the user's power distribution network in real time. The terminal's internal processing system then simultaneously calculates the three-phase active power and reactive power data. These three-phase voltage, current, active power, and reactive power data are referred to as basic electrical data. In this embodiment, the data acquisition frequency is once per second. In other embodiments, the implementer can set the acquisition frequency according to the actual implementation situation.
[0046] Furthermore, to eliminate calculation biases caused by different physical dimensions, normalization is performed on each type of basic electrical data. In this embodiment, the maximum-minimum value normalization algorithm is used. In other embodiments, implementers can select other normalization algorithms based on the distribution characteristics of each type of basic electrical data.
[0047] Since frequent adjustments to electrical load may degrade the physical operational stability of underlying circuit breakers and execution equipment, and the power change rate of most heavy electrical equipment is relatively slow, this invention sets up dynamic control of electrical load every certain time window. The most recent time window before each control trigger is taken as the current time period, and continuous basic electrical data within the most recent time window before each control trigger is extracted as the basic electrical data for the current time period. Simultaneously, considering that the production schedules of most factory transformer users typically exhibit a cyclical operation based on natural weeks, the week preceding the current time period is taken as the historical reference period, and basic electrical data within the week preceding the current time period is extracted as the basic electrical data for the historical reference period. This provides a basic reference system for subsequent analysis of the normal operating trajectory of equipment. In other embodiments, implementers can also set the length of the historical reference period according to the actual shift change physical cycle scheduled by the factory's manufacturing execution system.
[0048] The time window length is determined by the implementers based on the turnover cycle of the production batches within the target factory. In this embodiment, the empirical value of the time window length is 5 minutes, because most heavy machinery and equipment can complete the entire transient physical change process from power-on start-up to steady-state operation within 5 minutes. In other embodiments, the implementers can also set the time window length based on the average full-load start-up time of the electrical equipment.
[0049] S2: Determine the probability of total power anomaly and the probability of power anomaly in each phase based on the basic electrical data of the current time period, and thus determine whether the current time period is a suspected load anomaly period.
[0050] It should be noted that dedicated transformer users are not always under high load. During off-peak electricity consumption periods, dedicated transformer users do not require mandatory intervention from external terminals. Only when the internal power system approaches the physical limit of the distribution transformer capacity is intervention necessary to determine whether to increase the allowable power or limit the maximum power. Therefore, the total power risk performance and the evolution characteristics of power load imbalance in each phase within the current period determine the probability of load anomalies in the current period, which is used to determine whether the current distribution network is in a suspected period of load anomalies.
[0051] Specifically, the ratio of the total active power at each monitoring sampling moment in the current period to the maximum allowable total power set by the system is calculated to obtain the total power risk level at each monitoring sampling moment in the current period, satisfying the expression:
[0052]
[0053] in, Indicates the first time in the current time period The overall power risk level at each monitoring sampling time; Indicates the first time in the current time period Active power of phase A at each monitoring sampling time; Indicates the first time in the current time period The active power of phase B at each monitoring sampling time; Indicates the first time in the current time period The active power of phase C at each monitoring sampling time; This indicates the maximum allowable total power currently set by the system.
[0054] When certain large electromechanical equipment performs heavy-load startup operations, it causes the overall load of the power grid to exhibit a slow but continuous upward trend. Even if the transient total power value does not reach the extreme alarm line, this continuous accumulation of positive energy increments may still trigger the physical overload risk of thermal relay tripping. Therefore, this invention extracts the total power risk level between two adjacent monitoring times within the current time period and performs discrete difference calculations to obtain the total power trend risk level for the current time period.
[0055] Specifically, the risk level of the total power trend satisfies the expression:
[0056]
[0057] in, This indicates the degree of risk associated with the total power trend during the current period; This indicates the total number of monitoring and sampling times included in the current time period; Indicates the first time in the current time period The overall power risk level at each monitoring sampling time; Indicates the first time in the current time period The overall power risk level at each monitoring sampling time; This represents the normalization function. In this embodiment, a maximum-minimum normalization function is used. The global minimum extremum used in the calculation of the maximum-minimum normalization function is set to 0. The global maximum extremum is extracted from the average value of the difference in total power risk level between adjacent moments calculated from time windows of the same length as the current time period within the historical reference period. The global maximum value; in other embodiments, implementers can choose the normalization function according to the actual implementation situation. The difference in total power risk level between adjacent time points. When the total power remains constant and gradually increases, it indicates that the upward trend of the total power is more obvious, and there is a greater risk of overload. At this time, the risk level of the total power trend is greater.
[0058] Furthermore, since the total power risk level at the last monitoring sampling moment of the current period reflects the most realistic and urgent load pressure state of the current distribution network approaching its capacity limit in real time, the total power risk level at the last monitoring sampling moment of the current period is taken as the total power risk level of the current period. Based on the total power risk level and the total power trend risk level of the current period, the probability of total power anomalies in the current period is determined:
[0059]
[0060] in, This indicates the probability of an anomaly in total power during the current time period. This indicates the overall power risk level during the current period. This indicates the risk level of the total power trend in the current time period. If the risk level of the total power in the current time period... Surge or Total Power Trend Risk Level The persistently high levels indicate that the current power distribution network is under extremely high static overload pressure, or that its deterioration is progressing very rapidly, making overload collapse more likely.
[0061] Considering that some commercial buildings or mixed industrial parks have a large number of asymmetrical single-phase electrical devices installed inside their dedicated transformers, there is still a physical risk of overheating of single-phase cables even when the total power has not reached the upper limit. In order to lock the normal structural ratio of the three-phase load, the power distribution ratio of each phase at each historical sampling moment in the historical reference period is calculated. Specifically, for each historical sampling moment in the historical reference period, the sum of the active power of each phase at that historical sampling moment is taken as the total active power at that historical sampling moment, and the ratio between the active power of each phase at that historical sampling moment and the total active power is taken as the power distribution ratio of each phase at that historical sampling moment.
[0062] Using the three-phase power distribution ratio at each historical sampling time in the historical reference period as the first data sample, the density-based spatial clustering (DBSCAN) algorithm is used to cluster the first data sample to obtain multiple clusters. The three-phase power distribution ratio corresponding to the cluster center of the cluster containing the most samples is extracted as the normal benchmark ratio of three-phase power.
[0063] The product of the current system's maximum allowable total power and the normal power baseline ratio of each phase is used as the maximum limiting power of each phase. Based on the maximum limiting power of each phase and the active power of each phase at each monitoring sampling time in the current period, the power risk level of each phase at each monitoring sampling time in the current period is determined.
[0064]
[0065] in, Indicates the first time in the current time period The risk level of phase A power at each monitoring sampling time. Indicates the first time in the current time period The active power of phase A at each monitoring sampling time This indicates the maximum power limit for phase A.
[0066] Similarly, obtain the power risk level of phase B and phase C of each monitoring sample in the current time period.
[0067] Using the same method as for obtaining the overall power trend risk level for the current time period, the power trend risk level of each phase for the current time period is obtained. Specifically, the power risk level of phase A at two adjacent monitoring sampling times within the current time period is extracted and differentially calculated. Then, the arithmetic mean of the differences in the power risk levels of phase A at all adjacent times within the time period is normalized to obtain the power trend risk level of phase A, which characterizes the rate of deterioration of phase A load in the current time period. Similarly, the power trend risk levels of phase B and phase C for the current time period are obtained.
[0068] The power risk level of phase A at the last monitoring sampling moment in the current time period is taken as the power risk level of phase A in the current time period. The average of the power risk level of phase A in the current time period and the power trend risk level of phase A in the current time period is taken as the probability of power anomaly in phase A in the current time period. Similarly, the probabilities of power anomaly in phase B and phase C in the current time period are obtained.
[0069] Furthermore, since the overload risk may only occur in a certain phase, or each phase may be normal but the total power may be overloaded, the maximum value among the total power anomaly probability, the phase A power anomaly probability, the phase B power anomaly probability, and the phase C power anomaly probability in the current period is taken as the load anomaly probability in the current period.
[0070] If the probability of load anomaly in the current time period is greater than or equal to a preset anomaly warning threshold, the current time period is determined to be a suspected load anomaly period, requiring load regulation. The anomaly warning threshold is used to define the general decision-making baseline for when a single-phase or global multi-dimensional risk evolves to an uncontrollable state. In this embodiment, considering that when the comprehensive anomaly index approaches the 50% limit, the actual physical occurrence rate of electrical insulation damage and winding overheating in the power distribution network exhibits an exponential jump, the anomaly warning threshold is set to 0.5.
[0071] S3: In response to the current time period being a suspected load anomaly period, calculate the probability of each moment being an event point based on basic electrical data, and perform threshold segmentation to extract valid event points.
[0072] It should be noted that detected abnormal load spikes may originate from hard short-circuit faults such as insulation aging, or from the normal load start-up of core heavy-duty machine tools in the workshop. For anomalies caused by equipment failures, immediate load limiting and power outage actions must be implemented to protect the upstream power grid. However, if the anomaly is suspected to be caused by the start-up of core equipment, forced power limiting would cause a complete halt to the production line. The acceleration process of machine tools from standstill to operation is inevitably accompanied by a multi-dimensional, synchronous, and complex physical change, including a surge in active power, a synchronous rise in reactive power, and a short-term drop in the grid bus voltage due to a strong impact. This characteristic of mutual constraint and coordinated evolution of multiple physical quantities is the core physical difference distinguishing it from a simple conductor short-circuit fault. Since the physical manifestations of data coordination changes caused by the normal operation of different equipment are not entirely consistent, and the monitoring data fluctuations generated by the same equipment during its normal operation lifecycle usually have highly similar physical characteristics, this invention extracts the multi-dimensional physical changes and dynamic fluctuation immunity benchmarks of basic electrical data to pinpoint the key event points that induce real equipment operation events, providing precise time anchors for subsequent comparative analysis with historical data.
[0073] Specifically, for each type of basic electrical data, the difference between the basic electrical data at each monitoring sampling time in the current period and the basic electrical data at the previous time is taken as the change in basic electrical data at each monitoring sampling time in the current period. It should be noted that for the first monitoring sampling time in the current period, it is assumed that the basic electrical data has not changed, that is, the change in basic electrical data is 0.
[0074] During the high-load operation period of the factory's day shift, the background noise of the power grid itself is accompanied by significant natural harmonic fluctuations. Even when the equipment is in a stable operating phase, the basic electrical data will still exhibit continuous small jumps. To avoid misjudging the inherent harmonic noise as new equipment start-up and shutdown events, this invention extracts the arithmetic mean of the absolute values of the changes in basic electrical data within the previous noise baseline window for each monitoring sampling time in the current period. This arithmetic mean serves as the anti-interference benchmark for dynamic fluctuations, and based on this, the probability of each moment being an event point is calculated:
[0075]
[0076] in, Indicates the first time in the current time period The probability that a monitoring sampling time is an event point reflects the degree to which electrical fluctuations at the monitoring sampling time deviate from the recent stable background. Indicates the first time in the current time period The first monitoring sampling time Changes in basic electrical data; Indicates the total number of basic electrical data types; This indicates the extraction of the maximum value result from all types of basic electrical data; Indicates the absolute value symbol; Indicates the first Within the noise baseline window preceding the monitoring sampling time, the first The arithmetic mean of the absolute values of the changes in basic electrical data; This represents an overflow prevention constant, used to prevent the denominator from being 0. In this embodiment, it is... Setting it to 0.01, in other embodiments, the implementer can set this overflow prevention constant according to the factory-calibrated noise floor extreme value of the underlying current and voltage transformers. Since the start-up and shutdown actions of a single electromechanical device typically only cause a leapfrog change in some basic electrical data, this invention uses a maximum value extraction operation to perform extreme value screening of multi-dimensional features. The absolute value of the change of a certain basic electrical data at a specific monitoring sampling time. Far exceeding the recent average fluctuation At that time, the first The greater the likelihood that a monitoring sampling time is an event point.
[0077] The noise floor reference window is used to provide a smooth background scale that reflects the recent steady-state environment of the power grid. In this embodiment, considering that the transient impact process of the startup of dedicated transformer user equipment is mostly between a few seconds and tens of seconds, the length of the noise floor reference window is set to 0.5 seconds to ensure that the noise floor reference window can keenly capture the static noise floor characteristics before the sudden change without being swallowed up by the real startup long wave. In other embodiments, the implementer can set the length of the noise floor reference window according to the average switching electromagnetic transient delay of the reactive power compensation capacitor cabinet in the distribution network.
[0078] Furthermore, the probability of all monitoring sampling times within the current time period being event points is constructed into an event point probability sequence. The Otsu's method is used to perform threshold segmentation on this sequence, extracting the optimal cutoff threshold. When the probability of a monitoring sampling time being an event point is greater than or equal to this optimal cutoff threshold, that monitoring sampling time is determined to be a valid event point that has actually been triggered. If multiple adjacent monitoring sampling times that are consecutively determined to be event points exist on the timeline, only the monitoring sampling time with the earliest time series is retained as a valid event point, thereby eliminating multiple tailing misjudgments caused by a single physical activation action.
[0079] At this point, all valid event points within the current time period have been obtained.
[0080] Similarly, obtain all valid event points within the historical reference period.
[0081] S4: Divide the event segments using valid event points and merge them to construct a long historical sequence interval. Search for suspected similar historical windows in the long historical sequence interval to calculate the probability that the device corresponding to the current event segment is operating normally.
[0082] Based on the valid event points within the current time period, the current time period is divided into multiple event segments. Specifically: if no valid event point exists within the current time period, the current time period may be a continuation of a previous event, and is thus considered as one event segment; if valid event points exist within the current time period, the current time period is divided into multiple consecutive independent event segments, using each valid event point as a dividing point. It is important to note that a valid event point represents the physical switching instant when the equipment starts up from a standstill or transitions from one stable load state to another. Therefore, each valid event point necessarily serves as the starting point of a new physical state in the time sequence and is sequentially assigned to the next event segment directly triggered by that transition. The last event segment obtained from the current time period is taken as the current event segment.
[0083] Simultaneously, to establish a complete equipment condition comparison archive within historical long-term data, it is necessary to divide the historical reference period into event segments. Specifically: since the historical reference period and the current period are independent and do not overlap on the timeline, to prevent boundary data from failing to close the loop due to forced breaks in the timeline, the historical reference period and the periods preceding the current event segment in the current period are sequentially concatenated on the timeline to merge them into a continuous historical long-sequence interval. Subsequently, all valid event points within the historical long-sequence interval are used as dividing points to divide the historical long-sequence interval into multiple event segments, and each event segment within the historical long-sequence interval is designated as a historical event segment.
[0084] Obtain the basic electrical data type corresponding to the maximum value of various basic electrical data changes at the first monitoring sampling time in the current event segment, and use it as the target monitoring data. Establish a historical sliding window of the same length as the current event segment, and slide it through the historical long sequence interval with a sliding step size of 1. Calculate the similarity of the changing trends of each historical sliding window and the target monitoring data of the current event segment during the sliding process.
[0085]
[0086] in, Represents the first [number] in a long historical sequence interval. Similarity of target monitoring data trends between historical sliding windows and the current event segment; Represents an exponential function with the natural constant as its base; This indicates the total number of monitoring sampling moments included in the current event segment; Indicates the first event in the current event segment The change in target monitoring data at each monitoring sampling time; Represents the historical long sequence interval. The first in the history sliding window The change in target monitoring data at each monitoring sampling time. The change in target monitoring data within the historical sliding window. Change in target monitoring data compared to the current event segment When the waveforms of two waveforms become increasingly similar, the absolute value of their difference approaches 0, indicating a high degree of similarity in their trends. Approaching 1.
[0087] Furthermore, the similarity of the changing trends of target monitoring data between each historical sliding window in the historical long sequence interval and the current event segment is used as a second data sample. The second data sample is then clustered using the k-means clustering method to obtain multiple categories. The number of categories in the k-means clustering is determined using the elbow method. The maximum value of the changing trend similarity corresponding to the cluster centers of all categories is obtained, and each historical sliding window in the category corresponding to the maximum value is used as a suspected similar historical window for the current event segment.
[0088] Using the same calculation method as the target monitoring data's trend similarity, the similarity of the trend of various basic electrical data between each suspected similar historical window and the current event segment is obtained. Specifically, for each type of basic electrical data other than the target monitoring data, the sum of the absolute values of the differences between the change amount of that type of basic electrical data in each suspected similar historical window and the change amount of that type of basic electrical data at the corresponding time in the current event segment is calculated, and a smoothing mapping operation is performed using an exponential decay function with the natural constant as the base, thereby obtaining the similarity of the trend of the change of that type of basic electrical data between each suspected similar historical window and the current event segment.
[0089] Since events such as equipment startup may only cause significant changes in some basic electrical data, such as single-phase equipment having a smaller impact on other phases, this invention determines the weight of each basic electrical data's influence on the normal operation of the equipment in the current event segment based on the degree of change of each basic electrical data at the start of the current event segment.
[0090]
[0091] in, Indicates the first The weighting of the impact of basic electrical data on the normal operation of equipment in the current event segment; This represents the normalization function. In this embodiment, the maximum-minimum normalization function is used. The global maximum and minimum extrema used in the calculation of the maximum-minimum normalization function are extracted from the original relative change amplitudes calculated from all historical event segments and the current event segment. The dataset constitutes a set of data, thereby ensuring that the degree of change in electrical characteristics caused by the current operation of the equipment is objectively comparable throughout the factory's historical operating records. In other embodiments, implementers can choose a normalization function according to the actual implementation situation. Indicates the starting time of the current event segment. Changes in basic electrical data; This indicates the first event in the previous noise baseline window preceding the start time of the current event segment. The arithmetic mean of the absolute values of the changes in basic electrical data; This represents an overflow prevention constant, used to prevent the denominator from being 0. In this embodiment, it is... In other embodiments, the implementer can set this anti-overflow constant to 0.01 based on the factory-calibrated noise floor extreme value of the underlying current and voltage transformer.
[0092] Based on the similarity of the changing trends of various basic electrical data between each suspected similar historical window and the current event segment, and the influence weight of various basic electrical data on the normal operation of equipment in the current event segment, the probability of normal operation of equipment in the current event segment is determined:
[0093]
[0094] in, Indicates the probability that the device is operating normally in the current event segment; This indicates the total number of potentially similar historical windows in the current event segment; Indicates the total number of basic electrical data types; Indicates the first The first suspected similar historical window and the current event segment Similarity in the changing trends of basic electrical data; Indicates the first The weight of the impact of basic electrical data on the normal operation of equipment in the current event segment.
[0095] S5: Identify similar historical event segments based on the proportion of suspected similar historical windows contained in each historical event segment to assess the overall importance of the equipment, and generate a load adjustment index in combination with the probability of normal equipment operation.
[0096] It should be noted that when the power grid faces an overall load approaching its limit, for normally operating equipment, if the equipment is not critical, its power should be limited first to control electricity consumption; while if the equipment is critical, its normal operation should be guaranteed first, that is, the power limit should be appropriately increased to ensure the continuity and safety of the factory's core production line. Therefore, this invention further analyzes the importance of the equipment.
[0097] Specifically, the more frequently the current event segment appears in the historical reference period and the longer its duration, the higher its importance. The probability of similarity between each historical event segment and the current event segment is calculated based on the proportion of potentially similar historical windows contained within that historical event segment to the total number of windows contained in that historical event segment.
[0098]
[0099] in, Indicates the first The possibility that a historical event segment is similar to the current event segment; Indicates the first The number of potentially similar historical windows contained within a single historical event segment. Indicates the first The total number of historical sliding windows contained within each historical event segment.
[0100] In response to Explanation of the first Within each historical event segment, more than half of the historical sliding windows are similar to the current event segment. At this point, the [number]th [event]... If multiple consecutive adjacent historical event segments are determined to be similar to the current event segment on the timeline, it indicates that the physical waveform may have been temporarily truncated due to the superposition of other transient equipment start-up and shutdown events during the occurrence and continuous operation of a core event. In this case, all consecutive adjacent historical event segments similar to the current event segment on the timeline are spliced together end-to-end to form an independent complete event segment, which is recorded as a complete similar event segment. If a historical event segment is similar to the current event segment, but the historical event segments before and after it are not similar to the current event segment, then the isolated historical event segment is directly recorded as a complete similar event segment. This process is repeated throughout the entire long historical sequence interval to obtain all complete similar event segments similar to the current event segment.
[0101] The time importance of the device corresponding to the current event segment is determined based on the total number of complete similar event segments and the average duration of complete similar event segments.
[0102]
[0103] in, This indicates the time importance of the device corresponding to the current event segment. This represents the total number of complete similar event segments in the current event segment. This indicates the total number of event segments contained in the current time period and the historical time period. This represents the average duration of all similar event segments in the current event segment; This indicates the maximum duration of all event segments included in the current and historical time periods.
[0104] Furthermore, the greater the impact of changes in the equipment corresponding to the current event segment on the overall system, the higher the importance of that equipment. The weight of the impact of each basic electrical data on the normal operation of the equipment in the current event segment reflects the degree of significance of the changes in the basic electrical data. The start-up and shutdown of large machine tools or high-power equipment not only causes a huge surge in active power, but also causes a severe transient drop in bus voltage due to a surge in reactive power. That is, the more types of electrical data involved that experience violent fluctuations, and the greater the step amplitude of a single electrical data item, the stronger the equipment's steady-state destructive force and system penetration on the entire local distribution network. Therefore, this invention calculates the importance of the system impact of the equipment corresponding to the current event segment based on the sum of the impact weights of the basic electrical data on the normal operation of the equipment in the current event segment:
[0105]
[0106] in, This indicates the system impact importance of the device corresponding to the current event segment. Indicates the total number of basic electrical data types. Indicates the first The weight of the impact of basic electrical data on the normal operation of equipment in the current event segment, i.e., the weight of the impact of the first type of basic electrical data on the normal operation of equipment in the current event segment. The degree of change in basic electrical data; This represents the normalization function. In this embodiment, the maximum-minimum normalization function is used. The global maximum and global minimum values used in the calculation of the maximum-minimum normalization function are extracted from the sum of the influence weights calculated from all historical event segments and the current event segment. The dataset is constructed to ensure that the system impact of the equipment is comparable in the factory's historical operating records. In other embodiments, implementers can choose a normalization function based on the actual implementation situation.
[0107] Furthermore, the time importance of the device corresponding to the current event segment is determined. The system impact importance of the device corresponding to the current event segment The average value between these values is used as the overall importance of the device corresponding to the current event segment.
[0108] The higher the overall importance of the equipment, the greater the power required. The load adjustment intent parameters are determined based on the overall importance of the equipment corresponding to the current event segment.
[0109]
[0110] in, This parameter indicates the load adjustment intention. The positive or negative sign of the load adjustment intention parameter reflects the direction of adjustment for the corresponding equipment in the current event segment: whether to increase or limit power. This indicates the overall importance of the device corresponding to the current event segment. This represents the minimum of the combined importance of devices across all historical event segments and the combined importance of devices within the current time segment. This represents the maximum of the overall importance of the devices corresponding to all historical event segments and the overall importance of the devices corresponding to the current time segment. Used to assign importance Normalization to Within the range. When When this indicates that the device is critical, the system tends to increase the power limit for that device. When this occurs, it indicates that the device is not critical, and the system tends to reduce the available power limit for that device. It should be noted that when... At this point, the overall importance of the equipment is in a neutral critical state. Based on the principle of conservative protection for power grid security, equipment in a critical state should be subject to a slight power limitation first. Simultaneously, to avoid issues arising during subsequent adjustments... This leads to regulation failure, when At that time, based on the minimum data precision lower limit of the underlying digital processor, the load adjustment intention parameter is adjusted to... .
[0111] Furthermore, based on the load adjustment intention parameters and the probability of normal operation of equipment during the current event segment, the load control index is determined:
[0112]
[0113] in, Indicates the load adjustment index. Indicates the probability that the device is operating normally in the current event segment. This indicates the load adjustment intention parameter. When the overall importance of the equipment is low, causing the load adjustment intention parameter to be negative, but the probability of the equipment operating normally during the current event segment remains high. At higher levels, The calculated load control index is relatively small. For negative numbers with small absolute values, the control system performs small-scale power limiting actions on the equipment to prevent excessive disconnection of normally operating non-critical equipment; when the overall importance of the equipment is low, causing the load adjustment intention parameter to be negative, and the probability of the equipment operating normally in the current event segment is low... When smaller, The calculated load control index is relatively large. If the absolute value is a large negative number, the control system will implement a significant power reduction action on the non-critical equipment suspected of having a fault; when the overall importance of the equipment is high, causing the load adjustment intention parameter to be positive, and the probability of the equipment operating normally during the current event segment is high... When the value is large, the calculated load control index For positive numbers with large absolute values, the control system provides a large proportion of the available power expansion for the critical equipment that is operating normally; when the overall importance of the equipment is high, causing the load adjustment intention parameter to be positive, but the probability of the equipment operating normally in the current event segment is low. When the load control index is relatively small, the calculated load control index is... A positive number with a small absolute value, approaching 0, indicates that the control system reduces the power expansion ratio when the equipment exhibits potential fault characteristics. This maintains the basic power supply to the equipment while limiting the expansion of the electrical fault range. This ensures that basic power limits are applied to equipment of low importance but operating normally, sufficient available power is provided to equipment of high importance and operating normally, strict power reduction constraints are implemented for equipment of low importance but suspected of having an electrical fault, and minor capacity expansion is performed for equipment of high importance but suspected of having an electrical fault to prevent further collapse and spread of the grid fault.
[0114] S6: Adjust the maximum allowable total power and the maximum limited power of each phase based on the load adjustment index, and synchronously send the adjusted power parameters to the underlying intelligent equipment to perform power load control.
[0115] Specifically, the total power limit is adjusted based on the load adjustment index:
[0116]
[0117] in, Indicates the degree of adjustment of maximum total power. This indicates the maximum allowable total power currently set by the system. This indicates the load adjustment index.
[0118] To prevent unlimited power increases from causing transformer overload damage or excessive power reduction from causing power outages and shutdowns of factory infrastructure, safety boundaries are set for adjustments to the maximum permissible total power. Specifically: the maximum safe carrying capacity specified on the nameplate of the underlying distribution transformer and the minimum required power to ensure continuous operation of the factory infrastructure are obtained. The difference between the maximum safe carrying capacity and the current system's maximum permissible total power is used as the maximum permissible expansion limit; the difference between the minimum required power and the current system's maximum permissible total power is used as the maximum permissible reduction limit. For example, if the current system's maximum permissible total power is 1000kW, the minimum required power to ensure continuous operation of the factory infrastructure is 900kW, and the maximum safe carrying capacity specified on the nameplate of the underlying distribution transformer is 1100kW, then the calculated maximum permissible reduction limit is... The maximum allowed expansion amount is At this point, the maximum total power adjustment is limited to Within the numerical range.
[0119] In response to the maximum total power adjustment being greater than the maximum allowable expansion limit, the final maximum total power adjustment is set to the maximum allowable expansion limit; in response to the maximum total power adjustment being less than the maximum allowable reduction limit, the final maximum total power adjustment is set to the maximum allowable reduction limit; in response to the maximum total power adjustment being between the maximum allowable reduction limit and the maximum allowable expansion limit, the final maximum total power adjustment is equal to the supplementary limit of the global maximum allowable total power, thereby effectively blocking control commands that exceed the safe operating range of the hardware and ensuring that the underlying electrical equipment always operates within the objective safe range.
[0120] Furthermore, the power limit for each phase is adjusted separately based on the probability of power anomalies in each phase:
[0121]
[0122] in, This indicates the maximum power limit after adjustment for phase A; This indicates the maximum power limit for phase A; Indicates the degree of adjustment of the maximum total power; This indicates the probability of an abnormal power output in phase A during the current time period; This indicates the probability of abnormal power in phase B during the current time period; This indicates the probability of an abnormal C-phase power during the current time period.
[0123] The dedicated transformer acquisition terminal synchronously sends the adjusted maximum power limits for phase A, phase B, and phase C to the underlying intelligent circuit breakers and relay protection devices of the power distribution network. During real-time power monitoring, the underlying intelligent circuit breakers perform continuous over-limit identification actions based on the maximum power limits of the three phases. In response to any single phase's real-time active power exceeding the corresponding maximum power limit, they immediately trigger hardware trip power-off protection or flexible current limiting suppression operation.
[0124] In the description of this specification, "multiple" or "several" means at least two, such as two, three or more, unless otherwise expressly and specifically defined.
[0125] While this specification has shown and described numerous embodiments of the invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will occur to those skilled in the art without departing from the spirit and essence of the invention. It should be understood that various alternatives to the embodiments of the invention described herein may be employed in the practice of this invention.
Claims
1. A dedicated transformer data acquisition terminal with intelligent power load control function, comprising a memory, a processor, and a computer program stored in the memory and running on the processor, characterized in that, When the processor executes the computer program, it performs the following steps: Acquire basic electrical data for the current time period and historical reference time periods. The basic electrical data includes three-phase voltage data, three-phase current data, three-phase active power data, and three-phase reactive power data. Determine the probability of total power anomaly and the probability of power anomaly in each phase based on the basic electrical data for the current time period, thereby determining whether the current time period is a suspected load anomaly period. In response to the current period being a suspected period of abnormal load, valid event points are extracted based on basic electrical data, and the current period is divided into multiple event segments using the valid event points in the current period. The last event segment obtained from the current period is taken as the current event segment. The historical reference period is spliced together with the period before the current event segment in the current period on the time axis to form a continuous long historical sequence interval, and the long historical sequence interval is divided into multiple historical event segments using effective event points; Search for suspected similar historical windows that match the current event segment within a long historical sequence interval, and calculate the probability that the device corresponding to the current event segment is operating normally; Based on the proportion of suspected similar historical windows contained in each historical event segment, identify historical event segments similar to the current event segment, and assess the overall importance of the equipment corresponding to the current event segment based on the distribution characteristics of similar historical event segments and the magnitude of changes in electrical data of the current event segment. A load adjustment index is generated by combining the overall importance of the equipment and the probability of normal operation. The maximum allowable total power and the maximum limit power of each phase are adjusted based on the load adjustment index and then sent to the underlying intelligent equipment to perform power load regulation.
2. A dedicated transformer data acquisition terminal with intelligent power load control function according to claim 1, characterized in that, The step of determining the probability of total power anomalies and the probability of power anomalies in each phase based on the basic electrical data of the current time period, and thereby determining whether the current time period is a suspected period of load anomalies, includes: For any monitoring sampling moment in the current time period, the ratio between the sum of the three-phase active power at that monitoring sampling moment and the maximum allowable total power set by the current system is taken as the total power risk level at that monitoring sampling moment; the total power risk levels of adjacent monitoring sampling moments in the current time period are subjected to discrete difference operation and normalization processing to obtain the total power trend risk level of the current time period; the average of the total power risk level of the last monitoring sampling moment in the current time period and the total power trend risk level of the current time period is taken as the total power anomaly probability of the current time period. Extract the cluster centers of the three-phase power distribution ratio within the historical reference period as the normal benchmark ratio of three-phase power, and combine them with the maximum allowable total power set by the current system to obtain the maximum limited power of each phase. For any phase, the ratio between the active power of the phase and the maximum limiting power of the phase at each monitoring sampling time within the current time period is taken as the power risk level of the phase at each monitoring sampling time within the current time period; discrete difference calculation and normalization are performed on the power risk levels of the phase at adjacent monitoring sampling times within the current time period to obtain the power trend risk level of the phase; the average value between the power risk level of the phase at the last monitoring sampling time of the current time period and the power trend risk level of the phase in the current time period is taken as the probability of power anomaly in the current time period. The maximum value among the total power anomaly probability and the power anomaly probability of each phase is taken as the load anomaly probability; in response to the load anomaly probability being greater than or equal to the anomaly warning threshold, the current period is determined to be a suspected load anomaly period.
3. A dedicated transformer data acquisition terminal with intelligent power load control function according to claim 2, characterized in that, The risk level of the total power trend satisfies the expression: ； in, This indicates the degree of risk associated with the total power trend during the current period; This indicates the total number of monitoring and sampling times included in the current time period; Indicates the first [number]th ...time period] in the current time period. The overall power risk level at each monitoring sampling time; Indicates the first [number]th ...time period] in the current time period. The overall power risk level at each monitoring sampling time; This represents the normalization function.
4. A dedicated transformer data acquisition terminal with intelligent power load control function according to claim 1, characterized in that, The extraction of valid event points based on basic electrical data includes: For each type of basic electrical data, the arithmetic mean of the absolute values of the changes in basic electrical data within the previous noise baseline window at each monitoring sampling time is extracted as the anti-interference baseline for dynamic fluctuations. Based on the difference between the changes in basic electrical data at each monitoring sampling time and the anti-interference baseline, the probability of each monitoring sampling time being an event point is obtained through multi-dimensional feature extreme value screening to form an event point probability sequence. The event point probability sequence is segmented using the maximum inter-class variance threshold segmentation method to extract the optimal truncation threshold. Monitoring sampling times with a probability greater than or equal to the optimal truncation threshold are determined as valid event points. In response to the existence of multiple adjacent valid event points on the time axis, only the monitoring sampling time with the earliest time series is retained as a valid event point.
5. A dedicated transformer data acquisition terminal with intelligent power load control function according to claim 4, characterized in that, The probability that each monitoring sampling time is an event point satisfies the expression: ； in, Indicates the first [number]th ...time period] in the current time period. The probability that a monitoring sampling time is an event point; Indicates the first [number]th ...time period] in the current time period. The first monitoring sampling time Changes in basic electrical data; Indicates the total number of basic electrical data types; This indicates the extraction of the maximum value result from all types of basic electrical data; Indicates the absolute value symbol; Indicates the first Within the noise baseline window preceding the monitoring sampling time, the first The arithmetic mean of the absolute values of the changes in basic electrical data; This represents the constant for preventing overflow.
6. A dedicated transformer data acquisition terminal with intelligent power load control function according to claim 1, characterized in that, The step of finding a potentially similar historical window that matches the current event segment within a long historical sequence interval, and calculating the probability that the device corresponding to the current event segment is operating normally, includes: The basic electrical data type corresponding to the maximum value of the change in various basic electrical data at the first monitoring sampling time in the current event segment is taken as the target monitoring data. A historical sliding window with the same length as the current event segment is established and traversed in the long historical sequence interval. The similarity of the change trend of the target monitoring data between each historical sliding window and the current event segment is calculated, and a clustering algorithm is used to extract suspected similar historical windows. Based on the degree of change of each basic electrical data at the beginning of the current event segment, the influence weight of each basic electrical data on the normal operation of the equipment in the current event segment is determined. Combining the similarity of the change trend of each basic electrical data between each suspected similar historical window and the current event segment and the influence weight, the probability of the normal operation of the corresponding equipment in the current event segment is calculated.
7. A dedicated transformer data acquisition terminal with intelligent power load control function according to claim 6, characterized in that, The probability that the device corresponding to the current event segment is operating normally satisfies the expression: ； in, This indicates the probability that the device corresponding to the current event segment is operating normally; This indicates the total number of potentially similar historical windows in the current event segment; Indicates the total number of basic electrical data types; Indicates the first The first suspected similar historical window and the current event segment Similarity in the changing trends of basic electrical data; Indicates the first The weight of the impact of basic electrical data on the normal operation of equipment in the current event segment.
8. A dedicated transformer data acquisition terminal with intelligent power load control function according to claim 6, characterized in that, The process involves identifying historical event segments similar to the current event segment based on the proportion of suspected similar historical windows within each historical event segment, and assessing the overall importance of the equipment corresponding to the current event segment based on the distribution characteristics of similar historical event segments and the magnitude of changes in the electrical data of the current event segment. This includes: The proportion of the number of suspected similar historical windows contained in each historical event segment within a long historical sequence interval to the total number of windows contained in that historical event segment is used as the probability that each historical event segment is similar to the current event segment. In response to the probability of similarity being greater than a set threshold, the corresponding historical event segment is determined to be similar to the current event segment, and consecutively adjacent similar historical event segments on the timeline are merged into complete similar event segments, while isolated similar historical event segments are directly recorded as complete similar event segments. Based on the total number of complete similar event segments and the average duration of complete similar event segments within the entire historical long sequence interval, the temporal importance of the equipment corresponding to the current event segment is determined; the sum of the influence weights of various basic electrical data on the normal operation of the equipment in the current event segment is calculated and normalized to obtain the system influence importance of the equipment corresponding to the current event segment; the mean between the temporal importance and the system influence importance is taken as the comprehensive importance of the equipment corresponding to the current event segment.
9. A dedicated transformer data acquisition terminal with intelligent power load control function according to claim 1, characterized in that, The process of generating a load adjustment index by combining the overall importance of the equipment with the probability of normal operation includes: The overall importance of the equipment is normalized to the range of [-1, 1] to obtain the load adjustment intention parameter; in response to the load adjustment intention parameter being negative, the product of the difference between 1 and the probability of the equipment operating normally in the current event segment and the load adjustment intention parameter is used as the load adjustment index; in response to the load adjustment intention parameter being positive, the product of the probability of the equipment operating normally in the current event segment and the load adjustment intention parameter is used as the load adjustment index.
10. A dedicated transformer data acquisition terminal with intelligent power load control function according to claim 1, characterized in that, The adjustment of the maximum permissible total power and the maximum limited power of each phase based on the load adjustment index includes: The product of the current system's maximum permissible total power and the load adjustment index is taken as the maximum total power adjustment level. The maximum safe carrying capacity specified on the nameplate of the distribution transformer and the minimum required power of the factory's basic equipment are combined to apply boundary restrictions to the maximum total power adjustment level in order to obtain the adjusted maximum permissible total power. Adjust the power limit for each phase separately based on the probability of power anomalies in each phase: ； in, This indicates the maximum power limit after phase A is adjusted; This indicates the maximum power limit for phase A; Indicates the degree of adjustment of the maximum total power; This indicates the probability of an abnormal power output in phase A during the current time period; This indicates the probability of abnormal power in phase B during the current time period; This indicates the probability of an abnormal C-phase power during the current time period.