A cloud edge collaboration-based virtual power plant big data scheduling method

Abstract

The present application relates to the technical field of virtual power plant scheduling, and particularly relates to a virtual power plant big data scheduling method based on cloud edge cooperation. The present application obtains historical scheduling events and equipment operation parameters through the cloud, constructs a historical deviation curve of each operation parameter for a single device, intercepts an observation interval in the historical deviation curve, extracts a gravity position sequence of a qualified curve segment based on an actual operation state curve in the observation interval, calculates a gravity accumulation characteristic factor, determines a trend synchronization characteristic factor based on the gravity position sequence and the change trend of the same direction curve segment, analyzes the isomorphic characteristics of a single operation parameter, calibrates the isomorphic parameter, and then determines whether the device has an isomorphic covering tendency based on the isomorphic quantity proportion of the isomorphic parameter, so as to realize the identification of the early degradation of the device concealment, make up the technical blind area of the traditional single parameter threshold alarm, and improve the forward-looking and operation reliability of the virtual power plant scheduling decision under multiple uncertain factors.

Description

Technical Field

[0001] This invention relates to the field of virtual power plant scheduling technology, and in particular to a cloud-edge collaborative virtual power plant big data scheduling method. Background Technology

[0002] As the scale of virtual power plant aggregation and the complexity of cloud-edge collaborative scheduling continue to increase, the operational reliability of distributed energy equipment is not only closely related to the real-time over-limit monitoring of individual operating parameters, but also closely linked to the collaborative evolution of performance degradation of multiple devices and multiple operating parameters during long-term scheduling. Early identification of the "homogeneous masking" tendency of continuous deterioration of multi-parameter collaboration under the guise of compliance has become an indispensable part of achieving proactive preventive capacity regulation and precise operation and maintenance decisions.

[0003] Chinese Patent Publication No. CN117649090A discloses a virtual power plant scheduling optimization method and system, comprising: acquiring the type characteristics and spatiotemporal characteristics of energy equipment in the virtual power plant; acquiring the type characteristics and energy storage characteristics of energy storage equipment in the virtual power plant; collecting the electricity consumption of load equipment in the virtual power plant during the monitoring period, and predicting the electricity consumption of load equipment at the next moment based on the electricity consumption of load equipment in the virtual power plant; matching energy equipment to load equipment based on the electricity consumption of load equipment at the next moment, as well as the type characteristics, spatiotemporal characteristics, type characteristics, and energy storage characteristics of energy equipment. This application can efficiently integrate and schedule dispersed resources, thereby optimizing the scheduling of virtual power plants, enabling these energy equipment, energy storage equipment, and load equipment to work together better, improving energy utilization efficiency, and reducing energy waste.

[0004] However, the following problems still exist in the existing technology. In cloud-edge collaborative scheduling of virtual power plants, the normal response of equipment triggered by scheduling commands and the abnormal symptoms of early physical faults tend to be similar at the signal level. This may cause the system to mislabel fault signals as normal scheduling responses and exclude them, thereby reducing the sensitivity of equipment health status perception and the long-term operational safety of scheduling decisions. Summary of the Invention

[0005] To address this issue, the present invention provides a cloud-edge collaborative virtual power plant big data scheduling method to overcome the problem in the prior art that equipment fault signals may be mislabeled as normal scheduling responses and excluded, thereby reducing the sensitivity of equipment health status perception and the long-term operational security of scheduling decisions.

[0006] To achieve the above objectives, this invention provides a virtual power plant big data scheduling method based on cloud-edge collaboration, comprising: Step S1: Obtain from the cloud several historical scheduling events of the target area within a historical period, and the operating parameters of several devices corresponding to the historical scheduling events; Step S2: For a single device, construct the historical scheduling response curve corresponding to each operating parameter, and compare it with the preset benchmark response value to obtain several historical deviation curves. Based on the changing trend of the same-direction curve segments of each historical deviation curve, extract the observation interval of each operating parameter. Step S3: Within the observation interval, analyze the isomorphic features of a single operating parameter based on the centroid accumulation feature factor and the trend synchronization feature factor to calibrate the isomorphic parameter. Determine whether the device has an isomorphic masking tendency based on the proportion of isomorphic quantities of the isomorphic parameter. The centroid accumulation feature factor is determined based on the fitting slope of the centroid position sequence constructed from historical scheduling events. The trend synchronization feature factor is determined based on the convergence state of the centroid position sequence and the same-direction curve segment. Step S4: Based on the determination result, adjust the scheduling capacity of each device in the target area in the next scheduling cycle, and send it to the edge nodes corresponding to each device; Step S5: Based on the centroid accumulation characteristic factors of devices with isomorphic camouflage tendency obtained again after adjustment, determine the optimal intervention period and store it in the cloud; The operational parameters are obtained by the cloud based on the actual operational status response values ​​of the device in the corresponding historical scheduling events. The operational parameters include response duration, power change rate, steady-state recovery time, and startup response delay time.

[0007] Furthermore, the process of constructing historical scheduling response curves corresponding to each operating parameter and comparing them with preset baseline response values ​​to obtain several historical deviation curves includes, Extract the historical scheduling count based on the historical scheduling events; Using the historical scheduling count as the horizontal axis and the values ​​of each operating parameter as the vertical axis, construct the historical scheduling response curve corresponding to each operating parameter; Using the historical scheduling count as the horizontal axis and the difference between the value of each operating parameter and the corresponding baseline response value as the vertical axis, several historical deviation curves are obtained.

[0008] Furthermore, the process of extracting the observation interval for each operating parameter based on the changing trend of the same-direction curve segments of each of the historical deviation curves includes, Along the horizontal axis of each of the historical deviation curves, if the sign of the difference remains consistent over a predetermined number of scheduling times, then the starting number of scheduling times and the ending number of scheduling times with consistent signs are used as boundaries, and the curve segments within the boundaries are defined as curve segments in the same direction. For each historical deviation curve, if the same-direction curve segment shows a monotonically increasing trend with the scheduling events, then the number of scheduling events covered by the same-direction curve segment is determined as the observation interval.

[0009] Furthermore, the process of analyzing the isomorphic features of a single operating parameter based on the centroid accumulation feature factor and the trend synchronization feature factor to calibrate the isomorphic parameter includes, The isomorphic feature is obtained by weighted summation of the centroid accumulation feature factor and the trend synchronization feature factor; If the isomorphic feature is greater than or equal to the preset isomorphic threshold, then the running parameter is calibrated as an isomorphic parameter.

[0010] Furthermore, the process of determining whether the device has a tendency to mask similarities based on the proportion of similarities in the similarity parameter includes, The ratio of the number of the homomorphic parameters to the total number of the corresponding operating parameters of the device is determined as the homomorphic quantity ratio; If the proportion of the number of homomorphic objects is greater than or equal to a preset proportion threshold, it is determined that the device has a tendency to cover up homomorphic objects. If the proportion of the number of homomorphic shapes is less than a preset proportion threshold, it is determined that the device does not have a tendency to cover up homomorphic shapes.

[0011] Furthermore, the process of determining the centroid clustering feature factor by fitting the slope of the centroid position sequence constructed based on historical scheduling events includes, Based on the historical scheduling events, extract the numerical sequence of the actual operating status response of the device in each historical scheduling event; With time as the horizontal axis and the actual operating state response value sequence as the vertical axis, construct the actual operating state response curve corresponding to each historical scheduling event; From the actual operating state response curves, extract the curve segments corresponding to the vertical axis values ​​being within the preset target range, and determine several compliant curve segments. Determine the center point of the geometric area enclosed by each of the aforementioned standard curve segments and the corresponding horizontal axis, and determine the time coordinate corresponding to the center point as the centroid position; Arrange several centroid positions based on the number of historical scheduling times to construct a centroid position sequence; For a single operating parameter, a subsequence corresponding to the observation interval of the operating parameter is extracted from the centroid position sequence, the fitting slope of the subsequence is calculated, and it is determined as the centroid accumulation feature factor of the operating parameter.

[0012] Furthermore, the process of determining the trend synchronization characteristic factor based on the convergence state of the centroid position sequence and the same-direction curve segment includes, The observation interval is divided into several continuous sub-periods, and the centroid change trend of the centroid position sequence and the same-direction change trend of the length of the same-direction curve segment are determined in each sub-period. If it is determined that the trend of the change in the center of gravity is similar to the trend of change in the same direction, the sub-time period is marked; The ratio of the number of calibrated sub-time periods to the total number of sub-time periods within the observation interval is determined as the trend synchronization characteristic factor.

[0013] Furthermore, the process of determining whether the trend of change of the center of gravity is similar to the trend of change in the same direction includes, Linear fitting is performed on the corresponding centroid position sequence segments within the sub-time period, and the segment slope is determined; If the slope of the segment is positive, the trend of the center of gravity change is determined to be a backward shift of the center of gravity; if the slope of the segment is negative, the trend of the center of gravity change is determined to be a forward shift of the center of gravity. The number of scheduling times covered within the same-direction curve segment is determined as the length of the same-direction curve segment. The first length corresponding to the start end of the sub-period and the second length corresponding to the end end of the sub-period are calculated and determined. The difference between the second length and the first length is determined. If the difference is positive, the trend of change in the same direction is determined to be expansion in the same direction; if the difference is negative, the trend of change in the same direction is determined to be contraction in the same direction. If the shift of the center of gravity backward and the expansion in the same direction occur simultaneously, it is determined that the trend of the change in the center of gravity is similar to the trend of the change in the same direction.

[0014] Furthermore, the process of adjusting the scheduling capacity of each device within the target area in the next scheduling cycle based on the determination result includes, The scheduling capacity of devices with a tendency to mask shapes is removed and allocated to devices within the target area that do not have a tendency to mask shapes. The ratio of the scheduling capacity to its original scheduling capacity is positively correlated with the proportion of the number of devices with the tendency to mask each other.

[0015] Furthermore, the process of determining the optimal intervention period based on the centroid accumulation characteristic factor of the adjusted isomorphic camouflage-prone device includes, Identify the centroid accumulation characteristic factor after adjustment of equipment with isomorphic masking tendency; If the adjusted center of gravity accumulation characteristic factor is less than the preset center of gravity accumulation threshold for a continuous preset time length, then the end time of the preset time length is taken as the starting point, and a predetermined time domain is extended backward, and the predetermined time domain is taken as the optimal intervention period.

[0016] Compared with existing technologies, this invention obtains historical scheduling events and equipment operating parameters from the cloud, constructs observation intervals from the historical deviation curves of each operating parameter for individual equipment, extracts the centroid position sequence of the compliance curve segment based on the actual operating status curve of the equipment within the observation interval, and calculates the centroid accumulation feature factor. At the same time, it determines the trend synchronization feature factor based on the convergence of the centroid position sequence and the change of the same-direction curve segment, analyzes the isomorphic characteristics of individual operating parameters and calibrates the isomorphic parameters, and then determines whether the equipment has an isomorphic masking tendency based on the proportion of isomorphic parameters. This enables the early identification of the concealment of equipment, makes up for the technical blind spots of traditional single-parameter threshold alarms, and improves the foresight and operational reliability of virtual power plants in scheduling decisions under multiple uncertain factors.

[0017] In particular, this invention uses several historical deviation curves to extract the observation intervals of each operating parameter based on the changing trends of the same-direction curve segments of each historical deviation curve. In the practice of cloud-edge collaborative scheduling of virtual power plants, from the perspective of actual physical scenarios, whether it is the continuous decay of response rate caused by thermal aging of power devices in photovoltaic inverters, or the systematic extension of steady-state recovery time caused by intensified electrochemical polarization in energy storage systems, these real degradation processes cannot be exposed in one or two scheduling sessions, but accumulate gradually in continuous and directional scheduling cycles. The early performance degradation of operating parameters is often highly concealed: in long-cycle scheduling samples, degradation signals are not continuous, but sparsely and intermittently embedded in a large amount of normal response data. If all historical scheduling data is analyzed indiscriminately throughout the entire time period, it will not only cause huge cloud computing power overhead, but more importantly, it will confuse occasional normal fluctuations with those that truly have a continuous degradation trend, making it impossible to capture meaningful early warning signals before physical failure of equipment. Therefore, by using several historical deviation curves, a segment of the same curve along its horizontal axis is identified where the sign of the difference remains consistent across consecutive scheduling counts and exhibits a monotonically increasing trend. The range of scheduling counts covered by this segment is defined as the observation interval for this operating parameter. This observation interval essentially provides a precise calibration of the equipment degradation inertia window: it eliminates data interference from the equipment's healthy period and random fluctuation periods, focusing the analysis on the critical historical period during which the equipment's performance of this parameter is undergoing systematic and directional degradation. Within this interval, the degradation signal emerges from the masked background. The inertia of deviation expansion actively eliminates random deviation segments with uncertain directions and no monotonic properties, providing a reliable basis for subsequent analysis. This improves the accuracy of isomorphic masking tendency determination and the foresight of virtual power plant scheduling decisions.

[0018] In particular, this invention analyzes the isomorphic characteristics of individual operating parameters based on centroid accumulation feature factors and trend synchronization feature factors to calibrate isomorphic parameters. The proportion of isomorphic parameters determines whether the equipment exhibits a tendency for isomorphic masking. In the cloud-edge collaborative scheduling practice of virtual power plants, the numerical deviation of operating parameters in a sub-scheduling may originate from actual physical damage within the equipment, or it may be caused solely by external transient disturbances or short-term sensor drift. These two are highly similar in appearance, easily leading to misjudgment. Therefore, traditional threshold comparisons or single-point deviation analysis alone cannot distinguish between these two drastically different operating conditions. This application constructs the actual operating state response curve of the equipment in each historical scheduling event, extracts the compliant curve segments within the target range, and calculates the center time position of the geometric area enclosed by the compliant segment and the horizontal axis, thereby constructing a centroid position sequence reflecting the time structure of the equipment response process. The fitting slope of this sequence quantifies the accumulation rate of the centroid shifting backward along the time axis during the compliant period as the number of scheduling events increases. From a physical standpoint, this drift velocity is a direct reflection of the gradual accumulation of internal physical damage to the equipment. As thermal aging of power devices or wear of mechanical components intensifies, each response of the equipment requires more time to enter and maintain the target steady state. The overall center of gravity during the compliance period inevitably exhibits a systematic and directional backward shift. Therefore, the center of gravity accumulation characteristic factor can penetrate the normal appearance of operating parameter values ​​still within compliance thresholds and directly capture the continuous deterioration of the equipment's response time structure. Furthermore, in real industrial environments, actual physical degradation is a highly consistent process: it inevitably affects the equipment's response capability, leading to both a systematic backward shift of the center of gravity during the compliance period and a continuous, unidirectional expansion of the corresponding operating parameter deviations in multiple scheduling operations. Occasional disturbances do not exhibit this characteristic. For example, during a scheduling event, the deviation of operating parameters may show a significant expansion. However, instantaneous flicker in the external power grid voltage or short-term electromagnetic interference in the sensor sampling link could also cause similar amplitude fluctuations. If the actual operating state response curve of this scheduling event is extracted and the center of gravity of its compliance curve segment is calculated, it will be found that the center of gravity has not shifted synchronously. This indicates that the physical response process of the equipment has not truly deteriorated in terms of time structure, and the deviation is merely a false alarm at the data level. Conversely, when the equipment continues to deteriorate due to mechanical wear or thermal aging of power devices, the center of gravity of its compliance period has systematically shifted backward in consecutive scheduling events. However, due to the real-time compensation of the edge-side control strategy or the smoothing effect of the parameter calculation method, the operating parameter values ​​recorded during this period can still temporarily remain within the compliance range, and the apparent deviation has not yet significantly expanded, forming a typical masking state.Therefore, by dividing the observation interval into continuous sub-periods and examining whether the trend of the change in the center of gravity is similar to the trend of the change in the length of the same-direction curve segment, the proportion of similar sub-periods is statistically analyzed to obtain the trend synchronization characteristic factor. This factor filters out the two types of non-correlated interference mentioned above, eliminating false alarms where the deviation is large but the center of gravity has not moved, and identifying early cover-up where the center of gravity has moved but the deviation has not yet appeared. This ensures that only the coordinated decline of structural deterioration and apparent deterioration is labeled as the isomorphic parameter, thereby improving the accuracy of isomorphic cover-up tendency judgment. This provides early warning capability and foresight for the preventive capacity regulation and precise operation and maintenance decision-making of virtual power plants under the cloud-edge collaborative architecture.

[0019] In particular, this invention determines the optimal intervention period by adjusting the scheduling capacity of each device within the target area in the next scheduling cycle, based on the centroid accumulation characteristic factor of devices exhibiting isomorphic masking tendency after adjustment. In the cloud-edge collaborative scheduling practice of virtual power plants, when the cloud determines that a device exhibits isomorphic masking tendency, the device is actually already in the early decline stage of multi-parameter coordinated deterioration. Although its operating parameter values ​​have not yet triggered any over-limit alarms, damage to its internal physical structure is continuously accumulating. If the device is immediately forced out of operation, on the one hand, the real-time power balance of the regional power grid will be impacted due to the single-point capacity deficit; on the other hand, the opportunity window to continuously monitor the device's decline evolution during load reduction operation and provide decision-making basis for subsequent precise operation and maintenance will be missed. Since isomorphic masking tendency reveals the slow early decline of the device, the reasonable response strategy is not emergency shutdown, but rather to first reduce the scheduling capacity of the device with potential risks and transfer it to devices without isomorphic masking tendency, thereby mitigating the risk while ensuring the safe and stable operation of the power grid. Furthermore, relying solely on fixed cycles or experience to select the intervention time could lead to secondary damage due to premature intervention while the equipment is still in a state of deterioration inertia. Since the center-of-gravity accumulation characteristic factor is essentially a direct measure of the rate of deterioration in the equipment's time-to-standard structure, when the adjusted factor is less than a preset threshold over a sustained period, it indicates that the center-of-gravity shift in the equipment's response process is gradual, the degradation inertia is small, and the equipment's operating state has entered a relatively stable period. Therefore, this invention determines this sustained period as the optimal intervention period and stores it in the cloud, providing data-driven, precisely quantified optimal intervention time for operation and maintenance decisions, avoiding the high costs of blind shutdowns and reactive repairs in traditional operation and maintenance. This improves the operational reliability of the virtual power plant under conditions of uncertainty in equipment performance degradation. Attached Figure Description

[0020] Figure 1 This is a schematic diagram illustrating the steps of a cloud-edge collaborative virtual power plant big data scheduling method according to an embodiment of the invention. Figure 2 This is a logic decision diagram for calibrating the isomorphic parameters in an embodiment of the invention; Figure 3This is a logic block diagram illustrating whether a device for determining whether it has a tendency to cover up similar shapes, as described in an embodiment of the invention. Figure 4 A logic diagram for determining the optimal intervention period in an embodiment of the invention. Detailed Implementation

[0021] To make the objectives and advantages of the present invention clearer, the present invention will be further described below with reference to embodiments; it should be understood that the specific embodiments described herein are merely for explaining the present invention and are not intended to limit the present invention.

[0022] Preferred embodiments of the present invention will now be described with reference to the accompanying drawings. Those skilled in the art should understand that these embodiments are merely illustrative of the technical principles of the present invention and are not intended to limit the scope of protection of the present invention.

[0023] Furthermore, it should be noted that, in the description of this invention, unless otherwise explicitly specified and limited, the term "connection" should be interpreted broadly. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium; it can be a connection within two components. Those skilled in the art can understand the specific meaning of the above terms in this invention according to the specific circumstances.

[0024] Please see Figure 1 The diagram illustrates the steps of a cloud-edge collaborative virtual power plant big data scheduling method according to an embodiment of the invention. The cloud-edge collaborative virtual power plant big data scheduling method according to this embodiment includes: Step S1: Obtain from the cloud several historical scheduling events of the target area within a historical period, and the operating parameters of several devices corresponding to the historical scheduling events; Step S2: For a single device, construct the historical scheduling response curve corresponding to each operating parameter, and compare it with the preset benchmark response value to obtain several historical deviation curves. Based on the changing trend of the same-direction curve segments of each historical deviation curve, extract the observation interval of each operating parameter. Step S3: Within the observation interval, analyze the isomorphic features of a single operating parameter based on the centroid accumulation feature factor and the trend synchronization feature factor to calibrate the isomorphic parameter. Determine whether the device has an isomorphic masking tendency based on the proportion of isomorphic quantities of the isomorphic parameter. The centroid accumulation feature factor is determined based on the fitting slope of the centroid position sequence constructed from historical scheduling events. The trend synchronization feature factor is determined based on the convergence state of the centroid position sequence and the same-direction curve segment. Step S4: Based on the determination result, adjust the scheduling capacity of each device in the target area in the next scheduling cycle, and send it to the edge nodes corresponding to each device; Step S5: Based on the centroid accumulation characteristic factors of devices with isomorphic camouflage tendency obtained again after adjustment, determine the optimal intervention period and store it in the cloud; The operational parameters are obtained by the cloud based on the actual operational status response values ​​of the device in the corresponding historical scheduling events. The operational parameters include response duration, power change rate, steady-state recovery time, and startup response delay time.

[0025] In implementation, there are no restrictions on the method of obtaining a number of historical scheduling events of the target area within a historical period based on the cloud. They can be extracted directly from the historical operation database of the distributed devices through the cloud platform, or obtained by parsing the aggregated data packets periodically sent by the edge nodes. It is only necessary to obtain the device operation parameter values ​​and actual operation status value sequences corresponding to each historical scheduling event, which will not be elaborated further.

[0026] In practice, the historical period is usually selected within the range of [3 months, 12 months], and is preferably 6 months.

[0027] In implementation, the actual operating status response values ​​include, but are not limited to, one of the following: active power, frequency, voltage, and current that change continuously over time during a single scheduling instruction execution. These values ​​are used to construct the actual operating status response curves corresponding to each historical scheduling event, so as to extract the target curve segments and calculate the centroid position.

[0028] Specifically, the process of constructing historical scheduling response curves corresponding to each operating parameter and comparing them with preset baseline response values ​​to obtain several historical deviation curves includes, Extract the historical scheduling count based on the historical scheduling events; Using the historical scheduling count as the horizontal axis and the values ​​of each operating parameter as the vertical axis, construct the historical scheduling response curve corresponding to each operating parameter; Using the historical scheduling count as the horizontal axis and the difference between the value of each operating parameter and the corresponding baseline response value as the vertical axis, several historical deviation curves are obtained.

[0029] In practice, in the historical scheduling response curve, the vertical axis represents the recorded value of each operating parameter in each historical scheduling event, that is, the value calculated by the cloud based on the actual operating status response value sequence of the scheduling after the end of a single scheduling response of the equipment. The difference is the algebraic difference between the recorded value of each operating parameter and the corresponding preset benchmark response value.

[0030] In implementation, the purpose of the benchmark response value is to characterize the expected level of the operating parameters corresponding to the equipment under ideal health conditions, serving as an absolute reference benchmark for judging whether the operating parameters have deviated and the degree of deviation in each scheduling. The benchmark response value is predetermined. Those skilled in the art can select recorded values ​​of the operating parameter from several scheduling events in the early stages of equipment commissioning, for example, selecting recorded values ​​of steady-state recovery time from the first 20 historical scheduling events after equipment commissioning, and calculate their statistical mean to represent the baseline of the equipment's response capability under normal health conditions. To characterize the reasonable fluctuations allowed in actual equipment operation, the benchmark response value is set as the product of the statistical mean and the response error coefficient. Typically, the response error coefficient is selected within the range of [1.05, 1.20], and preferably 1.10 in implementation.

[0031] In implementation, there are no restrictions on the method of constructing historical scheduling response curves and historical deviation curves. They can be constructed using common data visualization tools or automatically generated by the data processing components built into the cloud platform. Simply map the running parameter values ​​of each scheduling and their differences from the baseline response values ​​to a two-dimensional coordinate system in the order of scheduling times to obtain the corresponding curves. This will not be elaborated further.

[0032] Specifically, the process of extracting the observation interval for each operating parameter based on the changing trend of the same-direction curve segments of each of the historical deviation curves includes, Along the horizontal axis of each of the historical deviation curves, if the sign of the difference remains consistent over a predetermined number of scheduling times, then the starting number of scheduling times and the ending number of scheduling times with consistent signs are used as boundaries, and the curve segments within the boundaries are defined as curve segments in the same direction. For each historical deviation curve, if the same-direction curve segment shows a monotonically increasing trend with the scheduling events, then the number of scheduling events covered by the same-direction curve segment is determined as the observation interval.

[0033] In practice, the predetermined number of scheduling times is usually selected within the range of [2 times, 5 times], and is preferably 3 times.

[0034] In implementation, the absolute value of the deviation can be smoothed using conventional data processing. If the absolute value of the deviation within the same-direction curve segment continues to increase with the number of scheduling operations, then the same-direction curve segment is determined to have a monotonically increasing trend.

[0035] In practice, each operating parameter has its own corresponding observation interval.

[0036] This invention uses several historical deviation curves to extract the observation intervals of each operating parameter based on the changing trends of the same-direction curve segments of each historical deviation curve. In the practice of cloud-edge collaborative scheduling of virtual power plants, from the perspective of actual physical scenarios, whether it is the continuous decay of response rate caused by thermal aging of power devices in photovoltaic inverters, or the systematic extension of steady-state recovery time due to intensified electrochemical polarization in energy storage systems, these real degradation processes cannot be exposed in one or two scheduling sessions, but accumulate gradually in continuous and directional scheduling cycles. The early performance degradation of operating parameters is often highly concealed: in long-cycle scheduling samples, degradation signals are not continuous, but sparsely and intermittently embedded in a large amount of normal response data. If all historical scheduling data is analyzed indiscriminately throughout the entire time period, it will not only cause huge cloud computing power overhead, but more importantly, it will confuse occasional normal fluctuations with those that truly have a continuous degradation trend, making it impossible to capture meaningful early warning signals before physical failure of equipment. Therefore, by using several historical deviation curves, a segment of the same curve along its horizontal axis is identified where the sign of the difference remains consistent across consecutive scheduling counts and exhibits a monotonically increasing trend. The range of scheduling counts covered by this segment is defined as the observation interval for this operating parameter. This observation interval essentially provides a precise calibration of the equipment degradation inertia window: it eliminates data interference from the equipment's healthy period and random fluctuation periods, focusing the analysis on the critical historical period during which the equipment's performance of this parameter is undergoing systematic and directional degradation. Within this interval, the degradation signal emerges from the masked background. The inertia of deviation expansion actively eliminates random deviation segments with uncertain directions and no monotonic properties, providing a reliable basis for subsequent analysis. This improves the accuracy of isomorphic masking tendency determination and the foresight of virtual power plant scheduling decisions.

[0037] Please see Figure 2 As shown in the diagram, the logical decision diagram for calibrating isomorphic parameters in an embodiment of the invention specifically includes the process of calibrating isomorphic parameters by analyzing the isomorphic features of a single operating parameter based on the centroid accumulation feature factor and the trend synchronization feature factor. The isomorphic feature is obtained by weighted summation of the centroid accumulation feature factor and the trend synchronization feature factor; If the isomorphic feature is greater than or equal to the preset isomorphic threshold, then the running parameter is calibrated as an isomorphic parameter.

[0038] In implementation, when weighted summing the centroid accumulation feature factor and the trend synchronization feature factor, the centroid accumulation feature factor is normalized to map it to a numerical range of 0 to 1. Then, the normalized centroid accumulation feature factor and the trend synchronization feature factor are weighted and summed to obtain the isomorphic feature. The method of normalizing the centroid accumulation feature factor is not limited. It can be achieved by linearly mapping the centroid accumulation feature factor by dividing it by the difference between the maximum and minimum values ​​of the centroid position sequence within the observation interval corresponding to the operating parameter, or by mapping it to the 0 to 1 range using the Sigmoid function. The only requirement is that the normalized centroid accumulation feature factor and the trend synchronization feature factor are additive in magnitude. The weight of the centroid accumulation feature factor is 0.55, and the weight of the trend synchronization feature factor is 0.45.

[0039] In practice, the purpose of the isomorphic threshold is to characterize the minimum comprehensive characteristic value required to determine whether a single operating parameter has an isomorphic deterioration pattern. Typically, the isomorphic threshold is selected within the range [0.5, 0.7], and is preferably 0.55 in practice.

[0040] Please see Figure 3 The diagram shown is a logic block diagram for determining whether a device has a tendency to cover up similar shapes, according to an embodiment of the invention. Specifically, the process of determining whether the device has a tendency to cover up similar shapes based on the proportion of similar shapes in the similarity parameter includes: The ratio of the number of the homomorphic parameters to the total number of the corresponding operating parameters of the device is determined as the homomorphic quantity ratio; If the proportion of the number of homomorphic objects is greater than or equal to a preset proportion threshold, it is determined that the device has a tendency to cover up homomorphic objects. If the proportion of the number of homomorphic shapes is less than a preset proportion threshold, it is determined that the device does not have a tendency to cover up homomorphic shapes.

[0041] In practice, the purpose of the percentage threshold is to characterize the minimum proportion of the same parameter required to determine whether the device has a multi-parameter collaborative systemic decline. Typically, the percentage threshold is selected within the range of [50%, 75%], and is preferably 50% in practice.

[0042] This invention analyzes the isomorphic characteristics of individual operating parameters based on centroid accumulation feature factors and trend synchronization feature factors to calibrate isomorphic parameters. The proportion of isomorphic parameters determines whether the equipment exhibits a tendency for isomorphic masking. In the cloud-edge collaborative scheduling practice of virtual power plants, the numerical deviation of operating parameters in a single scheduling operation may originate from actual physical damage within the equipment, or it may be caused solely by external transient disturbances or short-term sensor drift. These two are highly similar at the surface level, easily leading to misjudgment. Therefore, traditional threshold comparisons or single-point deviation analysis alone cannot distinguish between these two drastically different operating conditions. This application constructs the actual operating state response curve of the equipment in each historical scheduling event, extracts the compliant curve segments within the target range, and calculates the center time position of the geometric area enclosed by this compliant segment and the horizontal axis, thereby constructing a centroid position sequence reflecting the time structure of the equipment response process. The fitting slope of this sequence quantifies the accumulation rate of the centroid shifting backward along the time axis as the number of scheduling events increases during the compliant period. From a physical standpoint, this drift velocity is a direct reflection of the gradual accumulation of internal physical damage to the equipment. As thermal aging of power devices or wear of mechanical components intensifies, each response of the equipment requires more time to enter and maintain the target steady state. The overall center of gravity during the compliance period inevitably exhibits a systematic and directional backward shift. Therefore, the center of gravity accumulation characteristic factor can penetrate the normal appearance of operating parameter values ​​still within compliance thresholds and directly capture the continuous deterioration of the equipment's response time structure. Furthermore, in real industrial environments, actual physical degradation is a highly consistent process: it inevitably affects the equipment's response capability, leading to both a systematic backward shift of the center of gravity during the compliance period and a continuous, unidirectional expansion of the corresponding operating parameter deviations in multiple scheduling operations. Occasional disturbances do not exhibit this characteristic. For example, during a scheduling event, the deviation of operating parameters may show a significant expansion. However, instantaneous flicker in the external power grid voltage or short-term electromagnetic interference in the sensor sampling link could also cause similar amplitude fluctuations. If the actual operating state response curve of this scheduling event is extracted and the center of gravity of its compliance curve segment is calculated, it will be found that the center of gravity has not shifted synchronously. This indicates that the physical response process of the equipment has not truly deteriorated in terms of time structure, and the deviation is merely a false alarm at the data level. Conversely, when the equipment continues to deteriorate due to mechanical wear or thermal aging of power devices, the center of gravity of its compliance period has systematically shifted backward in consecutive scheduling events. However, due to the real-time compensation of the edge-side control strategy or the smoothing effect of the parameter calculation method, the operating parameter values ​​recorded during this period can still temporarily remain within the compliance range, and the apparent deviation has not yet significantly expanded, forming a typical masking state.Therefore, by dividing the observation interval into continuous sub-periods and examining whether the trend of the change in the center of gravity is similar to the trend of the change in the length of the same-direction curve segment, the proportion of similar sub-periods is statistically analyzed to obtain the trend synchronization characteristic factor. This factor filters out the two types of non-correlated interference mentioned above, eliminating false alarms where the deviation is large but the center of gravity has not moved, and identifying early cover-up where the center of gravity has moved but the deviation has not yet appeared. This ensures that only the coordinated decline of structural deterioration and apparent deterioration is labeled as the isomorphic parameter, thereby improving the accuracy of isomorphic cover-up tendency judgment. This provides early warning capability and foresight for the preventive capacity regulation and precise operation and maintenance decision-making of virtual power plants under the cloud-edge collaborative architecture.

[0043] Specifically, the process of determining the centroid clustering feature factor by fitting the slope of the centroid position sequence constructed based on historical scheduling events includes, Based on the historical scheduling events, extract the numerical sequence of the actual operating status response of the device in each historical scheduling event; With time as the horizontal axis and the actual operating state response value sequence as the vertical axis, construct the actual operating state response curve corresponding to each historical scheduling event; From the actual operating state response curves, extract the curve segments corresponding to the vertical axis values ​​being within the preset target range, and determine several compliant curve segments. Determine the center point of the geometric area enclosed by each of the aforementioned standard curve segments and the corresponding horizontal axis, and determine the time coordinate corresponding to the center point as the centroid position; Arrange several centroid positions based on the number of historical scheduling times to construct a centroid position sequence; For a single operating parameter, a subsequence corresponding to the observation interval of the operating parameter is extracted from the centroid position sequence, the fitting slope of the subsequence is calculated, and it is determined as the centroid accumulation feature factor of the operating parameter.

[0044] In implementation, suitable original electrical parameters can be selected from the actual operating state response numerical sequence based on the type of equipment being analyzed and its dispatch response characteristics. For example, for energy storage converters or photovoltaic inverters with active power regulation as the core dispatch objective, an active power time series can be selected as the actual operating state response numerical sequence; for primary frequency regulation units with frequency response as the core, a frequency time series can be selected; and for reactive power compensation devices with voltage regulation as the objective, a voltage time series can be selected. The selected original electrical parameters only need to accurately depict the continuous response process of the equipment under dispatch commands and have a clear target range for defining the target curve segment.

[0045] In implementation, the purpose of the target range is to characterize the allowable fluctuation range within which the actual operating state response value of the equipment in a single scheduling event should fall to be considered as meeting the standard. This range serves as the boundary for determining the segment of the curve that meets the standard from the actual operating state response curve. The target range is predetermined. Those skilled in the art can select the statistical mean of the actual operating state response value of the equipment in the steady-state phase from several scheduling events during the initial stage of equipment commissioning, for example, from the first 30 historical scheduling events after equipment commissioning, to represent the stable output level of the equipment after achieving the scheduling command target under normal conditions. To characterize the reasonable control error allowed in actual operation, the target range is set as the upper and lower bounds determined by a predetermined proportion of the statistical mean, i.e., the target range is [statistical mean × (1 - predetermined proportion), statistical mean × (1 + predetermined proportion)]. Typically, the predetermined proportion is selected within the range [0.02, 0.10], and preferably 0.07 in implementation.

[0046] In implementation, there is no limitation on the method of determining the center point of the geometric area. The benchmark curve segment can be discretized by numerical integration and its geometric centroid can be solved. Alternatively, the benchmark curve segment can be divided into several rectangular sub-regions according to a preset time step, and the average time coordinate of the area of ​​each rectangle can be used as the centroid position. As long as the center position of the geometric area enclosed by the benchmark curve segment and the horizontal axis can be accurately calculated on the time axis, this will not be elaborated further.

[0047] In implementation, when constructing the centroid position sequence based on the historical scheduling frequency, the sequence is arranged from smallest to largest, that is, from the earliest historical scheduling event to the latest historical scheduling event.

[0048] In implementation, there is no limitation on the method for calculating the fitting slope of the subsequence. Linear regression can be performed using the least squares method, or robust fitting can be performed using the Theil-Sen estimator. As long as the slope of the linear trend line of the centroid position changing with the number of scheduling times can be extracted from the centroid position point of the subsequence, it will not be elaborated further.

[0049] Specifically, the process of determining the trend synchronization characteristic factor based on the convergence state of the centroid position sequence and the same-direction curve segment includes, The observation interval is divided into several continuous sub-periods, and the centroid change trend of the centroid position sequence and the same-direction change trend of the length of the same-direction curve segment are determined in each sub-period. If it is determined that the trend of the change in the center of gravity is similar to the trend of change in the same direction, the sub-time period is marked; The ratio of the number of calibrated sub-time periods to the total number of sub-time periods within the observation interval is determined as the trend synchronization characteristic factor.

[0050] In implementation, there is no limitation on the method of dividing the observation interval into several consecutive sub-periods. Preferably, it can be divided at equal intervals according to a fixed number of scheduling. Typically, the number of scheduling in each sub-period is selected within the range of [3 times, 8 times], and in implementation, 5 times is preferred. If the total number of scheduling in the observation interval is not divisible by the length of the selected sub-period, the last sub-period is allowed to contain fewer scheduling times than the selected length, as long as it is not less than 3 times, it can be included in the trend synchronization determination. The last sub-period with fewer than 3 times is merged into the previous sub-period for processing.

[0051] Specifically, the process of determining whether the trend of change of the center of gravity is similar to the trend of change in the same direction includes, Linear fitting is performed on the corresponding centroid position sequence segments within the sub-time period, and the segment slope is determined; If the slope of the segment is positive, the trend of the center of gravity change is determined to be a backward shift of the center of gravity; if the slope of the segment is negative, the trend of the center of gravity change is determined to be a forward shift of the center of gravity. The number of scheduling times covered within the same-direction curve segment is determined as the length of the same-direction curve segment. The first length corresponding to the start end of the sub-period and the second length corresponding to the end end of the sub-period are calculated and determined. The difference between the second length and the first length is determined. If the difference is positive, the trend of change in the same direction is determined to be expansion in the same direction; if the difference is negative, the trend of change in the same direction is determined to be contraction in the same direction. If the shift of the center of gravity backward and the expansion in the same direction occur simultaneously, it is determined that the trend of the change in the center of gravity is similar to the trend of the change in the same direction.

[0052] In practice, there is no limitation on the method for calculating the slope of the segment. It can be obtained by linear fitting using the least squares method or the Theil-Sen estimator. It is only necessary to extract the slope of the local trend line of the change of the centroid position with the number of scheduling from the centroid position sequence segment. This will not be elaborated further.

[0053] In implementation, the first length is the length of the same-direction curve segment corresponding to the number of times the sub-period starts scheduling, that is, the number of scheduling times covered by the continuous same-direction deviation between the starting point of the same-direction curve segment and the number of times the sub-period starts scheduling; the second length is the length of the same-direction curve segment corresponding to the number of times the sub-period ends scheduling, that is, the number of scheduling times covered by the continuous same-direction deviation between the starting point of the same-direction curve segment and the number of times the sub-period ends scheduling.

[0054] In implementation, the simultaneous occurrence of the backward shift and the same-direction expansion of the center of gravity means that within a sub-period, the conditions of a positive local fitting slope for the center of gravity position sequence and a positive difference in the length of the same-direction curve segments are simultaneously met. Only under this condition is the trend of the center of gravity change determined to be convergent with the trend of the same-direction change. For the forward shift and the same-direction contraction, the backward shift and the same-direction contraction, the forward shift and the same-direction expansion, and any situation where the trend is flat, convergence is not considered. This is because the "convergence" defined in this invention has a specific technical orientation; its object is a coordinated deterioration event in the device response process across both the temporal structure and apparent numerical dimensions, rather than directional synchronization in the general sense. The simultaneous occurrence of the shift in the center of gravity and the expansion in the same direction indicates that the systematic shift in the center of gravity and the continuous expansion of the deviation of operating parameters during the period of compliance were observed at the same time window. Both of them point to a unified and continuous physical decay mechanism inside the equipment. Other combinations are either synergistic improvements or two dimensions that are contrary to each other. They do not constitute evidence of "masking decay" in a physical sense. Including them in the judgment will introduce noise interference and reduce the accuracy and indicativeness of the diagnosis of isomorphic masking tendency.

[0055] Specifically, the process of adjusting the scheduling capacity of each device in the target area in the next scheduling cycle based on the determination result includes, The scheduling capacity of devices with a tendency to mask shapes is removed and allocated to devices within the target area that do not have a tendency to mask shapes. The ratio of the scheduling capacity to its original scheduling capacity is positively correlated with the proportion of the number of devices with the tendency to mask each other.

[0056] In implementation, optional, If the proportion of the number of identical shapes is greater than or equal to the first proportion threshold, then the removal ratio is 30%. If the proportion of the number of identical shapes is less than the first proportion threshold and greater than the second proportion threshold, then the removal ratio is 20%. If the proportion of the same shape is less than or equal to the second proportion threshold, then the removal ratio is 10%. The first quantity percentage threshold is 1.25 times the preset percentage threshold, and the second quantity percentage threshold is 1.15 times the preset percentage threshold.

[0057] In practice, the original scheduling capacity is the scheduling capacity value that the cloud plans for the device in the next scheduling cycle when it determines that the device does not have a tendency to cover up the same shape. This value can be generated by the cloud based on the load forecast in the target area, the rated power of each device and historical scheduling records through a conventional optimization scheduling algorithm. It only needs to represent the scheduling capacity that the device should have undertaken when the same shape cover intervention is not triggered, which will not be elaborated further.

[0058] Please see Figure 4 As shown, this is a logic diagram for determining the optimal intervention period according to an embodiment of the invention. Specifically, the process of determining the optimal intervention period based on the centroid accumulation characteristic factor of the device with a tendency to cover up similar shapes after adjustment includes: Identify the centroid accumulation characteristic factor after adjustment of equipment with isomorphic masking tendency; If the adjusted center of gravity accumulation characteristic factor is less than the preset center of gravity accumulation threshold for a continuous preset time length, then the end time of the preset time length is taken as the starting point, and a predetermined time domain is extended backward, and the predetermined time domain is taken as the optimal intervention period.

[0059] In practice, the method for determining the centroid accumulation characteristic factor of equipment with homomorphic masking tendency after adjustment can refer to the aforementioned calculation method of centroid accumulation characteristic factor. Based on multiple consecutive historical scheduling events newly collected after capacity adjustment, the centroid position sequence is reconstructed and its fitting slope is calculated. This slope reflects the changing trend of the centroid of the equipment after adjustment to meet the standard. This will not be elaborated further.

[0060] In practice, the preset time length is usually selected within the range of [24 hours, 72 hours], and is preferably 48 hours.

[0061] In practice, the center of gravity accumulation threshold is a positive number approaching zero, typically selected within the range [0.05, 0.15], and preferably 0.10. When the adjusted center of gravity accumulation factor is less than this threshold, it indicates that the systematic backward shift of the target center of gravity has been effectively curbed, and the equipment operation has entered a relatively stable period.

[0062] In practice, the predetermined time domain is usually selected within the interval [8 hours, 28 hours], and is preferably 24 hours.

[0063] This invention determines the optimal intervention period by adjusting the scheduling capacity of each device within the target area in the next scheduling cycle, based on the centroid accumulation characteristic factor of devices exhibiting isomorphic masking tendency after adjustment. In the cloud-edge collaborative scheduling practice of virtual power plants, when the cloud determines that a device exhibits isomorphic masking tendency, the device is actually already in the early decline stage of multi-parameter coordinated deterioration. Although its operating parameter values ​​have not yet triggered any over-limit alarms, damage to its internal physical structure is continuously accumulating. If the device is immediately forced out of operation, on the one hand, the real-time power balance of the regional power grid will be impacted due to the single-point capacity deficit; on the other hand, the opportunity window for continuously monitoring the device's decline evolution during load reduction operation and providing decision-making basis for subsequent precise operation and maintenance will be missed. Since isomorphic masking tendency reveals the slow early decline of the device, the reasonable response strategy is not emergency shutdown, but rather to first reduce the scheduling capacity of the device with potential risks and transfer it to devices without isomorphic masking tendency, thereby mitigating the risk while ensuring the safe and stable operation of the power grid. Furthermore, relying solely on fixed cycles or experience to select the intervention time could lead to secondary damage due to premature intervention while the equipment is still in a state of deterioration inertia. Since the center-of-gravity accumulation characteristic factor is essentially a direct measure of the rate of deterioration in the equipment's time-to-standard structure, when the adjusted factor is less than a preset threshold over a sustained period, it indicates that the center-of-gravity shift in the equipment's response process is gradual, the degradation inertia is small, and the equipment's operating state has entered a relatively stable period. Therefore, this invention determines this sustained period as the optimal intervention period and stores it in the cloud, providing data-driven, precisely quantified optimal intervention time for operation and maintenance decisions, avoiding the high costs of blind shutdowns and reactive repairs in traditional operation and maintenance. This improves the operational reliability of the virtual power plant under conditions of uncertainty in equipment performance degradation.

[0064] The technical solution of the present invention has been described above with reference to the preferred embodiments shown in the accompanying drawings. However, it will be readily understood by those skilled in the art that the scope of protection of the present invention is obviously not limited to these specific embodiments. Without departing from the principles of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical features, and the technical solutions after these changes or substitutions will all fall within the scope of protection of the present invention.

Claims

1. A virtual power plant big data scheduling method based on cloud-edge collaboration, characterized in that, include, Step S1: Obtain from the cloud several historical scheduling events of the target area within a historical period, and the operating parameters of several devices corresponding to the historical scheduling events; Step S2: For a single device, construct the historical scheduling response curve corresponding to each operating parameter, and compare it with the preset benchmark response value to obtain several historical deviation curves. Based on the changing trend of the same-direction curve segments of each historical deviation curve, extract the observation interval of each operating parameter. Step S3: Within the observation interval, analyze the isomorphic features of a single operating parameter based on the centroid accumulation feature factor and the trend synchronization feature factor to calibrate the isomorphic parameter. Determine whether the device has an isomorphic masking tendency based on the proportion of isomorphic quantities of the isomorphic parameter. The centroid accumulation feature factor is determined based on the fitting slope of the centroid position sequence constructed from historical scheduling events. The trend synchronization feature factor is determined based on the convergence state of the centroid position sequence and the same-direction curve segment. Step S4: Based on the determination result, adjust the scheduling capacity of each device in the target area in the next scheduling cycle, and send it to the edge nodes corresponding to each device; Step S5: Based on the centroid accumulation characteristic factors of devices with isomorphic camouflage tendency obtained again after adjustment, determine the optimal intervention period and store it in the cloud; The operational parameters are obtained by the cloud based on the actual operational status response values ​​of the device in the corresponding historical scheduling events. The operational parameters include response duration, power change rate, steady-state recovery time, and startup response delay time.

2. The cloud-edge collaborative virtual power plant big data scheduling method according to claim 1, characterized in that, The process of constructing historical scheduling response curves corresponding to each operating parameter and comparing them with preset baseline response values ​​to obtain several historical deviation curves includes: Extract the historical scheduling count based on the historical scheduling events; Using the historical scheduling count as the horizontal axis and the values ​​of each operating parameter as the vertical axis, construct the historical scheduling response curve corresponding to each operating parameter; Using the historical scheduling count as the horizontal axis and the difference between the value of each operating parameter and the corresponding baseline response value as the vertical axis, several historical deviation curves are obtained.

3. The cloud-edge collaborative virtual power plant big data scheduling method according to claim 2, characterized in that, The process of extracting the observation interval for each operating parameter based on the changing trend of the same-direction curve segments of each of the historical deviation curves includes: Along the horizontal axis of each of the historical deviation curves, if the sign of the difference remains consistent over a predetermined number of scheduling times, then the starting number of scheduling times and the ending number of scheduling times with consistent signs are used as boundaries, and the curve segments within the boundaries are defined as curve segments in the same direction. For each historical deviation curve, if the same-direction curve segment shows a monotonically increasing trend with the scheduling events, then the number of scheduling events covered by the same-direction curve segment is determined as the observation interval.

4. The cloud-edge collaborative virtual power plant big data scheduling method according to claim 1, characterized in that, The process of analyzing the isomorphic features of a single operating parameter based on the centroid accumulation feature factor and the trend synchronization feature factor, in order to calibrate the isomorphic parameter, includes the following: The isomorphic feature is obtained by weighted summation of the centroid accumulation feature factor and the trend synchronization feature factor; If the isomorphic feature is greater than or equal to the preset isomorphic threshold, then the running parameter is calibrated as an isomorphic parameter.

5. The virtual power plant big data scheduling method based on cloud-edge collaboration according to claim 1, characterized in that, The process of determining whether the device has a tendency to mask similar shapes based on the proportion of similar shapes in the similarity parameter includes: The ratio of the number of the homomorphic parameters to the total number of the corresponding operating parameters of the device is determined as the homomorphic quantity ratio; If the proportion of the number of homomorphic objects is greater than or equal to a preset proportion threshold, it is determined that the device has a tendency to cover up homomorphic objects. If the proportion of the number of homomorphic shapes is less than a preset proportion threshold, it is determined that the device does not have a tendency to cover up homomorphic shapes.

6. The cloud-edge collaborative virtual power plant big data scheduling method according to claim 1, characterized in that, The process of determining the centroid clustering feature factor by fitting the slope of the centroid position sequence constructed based on historical scheduling events includes, Based on the historical scheduling events, extract the numerical sequence of the actual operating status response of the device in each historical scheduling event; With time as the horizontal axis and the actual operating state response value sequence as the vertical axis, construct the actual operating state response curve corresponding to each historical scheduling event; From the actual operating state response curves, extract the curve segments corresponding to the vertical axis values ​​being within the preset target range, and determine several compliant curve segments. Determine the center point of the geometric area enclosed by each of the aforementioned standard curve segments and the corresponding horizontal axis, and determine the time coordinate corresponding to the center point as the centroid position; Arrange several centroid positions based on the number of historical scheduling times to construct a centroid position sequence; For a single operating parameter, a subsequence corresponding to the observation interval of the operating parameter is extracted from the centroid position sequence, the fitting slope of the subsequence is calculated, and it is determined as the centroid accumulation feature factor of the operating parameter.

7. The virtual power plant big data scheduling method based on cloud-edge collaboration according to claim 1, characterized in that, The process of determining the trend synchronization characteristic factor based on the convergence state of the centroid position sequence and the same-direction curve segment includes, The observation interval is divided into several continuous sub-periods, and the centroid change trend of the centroid position sequence and the same-direction change trend of the length of the same-direction curve segment are determined in each sub-period. If it is determined that the trend of the change in the center of gravity is similar to the trend of change in the same direction, the sub-time period is marked; The ratio of the number of calibrated sub-time periods to the total number of sub-time periods within the observation interval is determined as the trend synchronization characteristic factor.

8. The virtual power plant big data scheduling method based on cloud-edge collaboration according to claim 7, characterized in that, The process of determining whether the trend of the center of gravity change is similar to the trend of change in the same direction includes: Linear fitting is performed on the corresponding centroid position sequence segments within the sub-time period, and the segment slope is determined; If the slope of the segment is positive, the trend of the center of gravity change is determined to be a backward shift of the center of gravity; if the slope of the segment is negative, the trend of the center of gravity change is determined to be a forward shift of the center of gravity. The number of scheduling times covered within the same-direction curve segment is determined as the length of the same-direction curve segment. The first length corresponding to the start end of the sub-period and the second length corresponding to the end end of the sub-period are calculated and determined. The difference between the second length and the first length is determined. If the difference is positive, the trend of change in the same direction is determined to be expansion in the same direction; if the difference is negative, the trend of change in the same direction is determined to be contraction in the same direction. If the shift of the center of gravity backward and the expansion in the same direction occur simultaneously, it is determined that the trend of the change in the center of gravity is similar to the trend of the change in the same direction.

9. The cloud-edge collaborative virtual power plant big data scheduling method according to claim 1, characterized in that, The process of adjusting the scheduling capacity of each device in the target area in the next scheduling cycle based on the determination result includes, The scheduling capacity of devices with a tendency to mask shapes is removed and allocated to devices within the target area that do not have a tendency to mask shapes. The ratio of the scheduling capacity to its original scheduling capacity is positively correlated with the proportion of the number of devices with the tendency to mask each other.

10. The cloud-edge collaborative virtual power plant big data scheduling method according to claim 9, characterized in that, The process of determining the optimal intervention period based on the centroid accumulation characteristic factor of the adjusted device with a tendency to cover up similar shapes includes: Identify the centroid accumulation characteristic factor after adjustment of equipment with isomorphic masking tendency; If the adjusted center of gravity accumulation characteristic factor is less than the preset center of gravity accumulation threshold for a continuous preset time length, then the end time of the preset time length is taken as the starting point, and a predetermined time domain is extended backward, and the predetermined time domain is taken as the optimal intervention period.

Images