A management method for power equipment

Abstract

The present application discloses a management method for power equipment. The method includes: monitoring the power of each power equipment in a power scenario in real time to obtain a load change response feature vector corresponding to the power equipment when a load mutation condition is satisfied; based on each power equipment, inputting the load change response feature vector into a pre-trained power equipment power adjustment behavior prediction model, and outputting adjustment behavior information corresponding to the power equipment, including a target power adjustment amount, a response time, and a charging and discharging rate matching of a buffer device; based on the adjustment behavior information, using a pre-set progressive adjustment algorithm to generate scheduling instruction information for the power equipment, so as to achieve reasonable scheduling of the power equipment. Thereby, equipment loss is reduced, and the stability of power equipment management in a power system is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of power equipment management, and particularly to a management method for power equipment. Background Art

[0002] Power equipment mainly includes power generation equipment, power transmission equipment, power distribution equipment and power consumption equipment. Power generation equipment usually includes generators, wind turbine generators, solar photovoltaic panels, etc. Power transmission equipment usually includes transformers, switchgear, cables, etc. Power distribution equipment usually includes switch cabinets, distribution boards, power meters, etc. Power consumption equipment usually includes motors, lighting equipment, air conditioning equipment, etc. Power equipment management refers to the process of effectively monitoring, controlling, maintaining and optimizing various equipment in the power system. Its purpose is to ensure the normal operation of power equipment, extend the equipment life, reduce the failure risk, and improve the reliability, safety and energy efficiency of the power system. At present, large voltage fluctuations are likely to occur during the peak and trough periods of power consumption of power equipment. At the same time, power needs to be cut off during the maintenance of power equipment, and this process will also cause large voltage fluctuations. And large voltage fluctuations may lead to power equipment failures or power outages, making it difficult to manage power equipment.

[0003] Chinese invention patent with the patent application number 202311656990.X discloses a power equipment management method, which performs dynamic simulation according to equipment information to generate power equipment image information and power buffer equipment image information, extracts operating condition data according to the image information, performs power matching, and obtains management information according to the power matching information and equipment load data, so as to implement the scheduling strategy of power equipment and ensure power stability and load balance.

[0004] However, in complex and changeable power scenarios, high-precision power equipment (such as precision manufacturing, medical equipment, etc.) is sensitive to the instantaneous response of load changes. For example, a certain industrial park has deployed photovoltaic power generation, energy storage battery packs and industrial load equipment. Since the photovoltaic output fluctuates violently due to weather effects (such as sudden voltage drops caused by cloud cover) and there are instantaneous power surges in industrial loads, these load fluctuations need to be smoothed by power buffer equipment (such as supercapacitors or flywheel energy storage) to ensure stable power supply. It can be seen that the coordinated operation of power equipment and buffer equipment is crucial for system stability; however, traditional abrupt scheduling (directly adjusting the equipment output power) directly executes the scheduling instruction according to the feedback management information, which is likely to cause the following problems: frequent start and stop of equipment, frequent switching of equipment caused by sudden load changes, accelerating mechanical wear; rapid load changes leading to violent fluctuations in electromagnetic radiation / thermal radiation of equipment, affecting equipment stability and the safety of the surrounding environment; voltage instantaneous fluctuations caused by mismatched charge and discharge rates of buffer equipment, reducing the reliability of the power grid; instantaneous power demand exceeding the rated capacity of the equipment, shortening the equipment life. Summary of the Invention

[0005] The present application provides a management method for power equipment, enabling the power equipment to make reasonable response scheduling during load mutation, reducing equipment loss, and improving the stability of power equipment management in the power system.

[0006] The present application provides a management method for power equipment, including:

[0007] S101, monitoring the power of each power equipment in the power scenario in real time, and obtaining the load change response feature vector corresponding to the power equipment when the load mutation condition is satisfied;

[0008] S102, based on each power equipment, inputting the load change response feature vector into a pre-trained power equipment power adjustment behavior prediction model, and outputting the adjustment behavior information corresponding to the power equipment, including the target power adjustment amount, response time, and charge and discharge rate matching of the buffer equipment;

[0009] S103, based on the adjustment behavior information, using a preset progressive adjustment algorithm to generate the scheduling instruction information of the power equipment, and realizing the reasonable scheduling of the power equipment.

[0010] Preferably, the load mutation condition is set as: the load change amount of the power equipment is greater than the preset load change threshold corresponding to the equipment type to which the power equipment belongs; wherein, different equipment types correspond to different load change thresholds, which are set according to the actual situation and expert experience and are used to reflect the mutation degree of the load change of the power equipment.

[0011] Preferably, the load change response feature vector is set as , where s is the coding value of the equipment type to which the power equipment belongs, P is the current power value of the power equipment, is the load step change amount, is the load change rate; R is the response feature, which is set as , ( )], where T is the current temperature of the power equipment, is the maximum power adjustment rate, ( ) is the efficiency decay coefficient.

[0012] Preferably, the pre-trained power equipment power adjustment behavior prediction model is obtained in the following way:

[0013] A1. Collect a large number of historical load change response feature vectors of power equipment that satisfy the corresponding load mutation conditions in history, and perform a preset screening mechanism to obtain the load change response feature vectors to be trained;

[0014] A2. Label the response feature vectors for the training load changes, and set the labeling content as adjustment behavior information, including the target power adjustment amount, response time, and charge and discharge rate matching of the buffer device;

[0015] A3. Use all the labeled load change response feature vectors as the training set to train and learn the pre-selected neural network structure, and continuously optimize the model parameters to generate the final power device power adjustment behavior prediction model.

[0016] Preferably, the preset screening mechanism is set as:

[0017] C1. Based on each historical load change response feature vector, calculate the power buffer matching value according to the following formula:

[0018]

[0019] where K is the power buffer matching value, is the maximum voltage value absorbed or released by the buffer device responding to the power device within the response time, is the voltage difference of the corresponding power device within the response time;

[0020] C2. Determine all historical load change response feature vectors with a power buffer matching value greater than the preset matching threshold as the load change response feature vectors to be trained.

[0021] Preferably, in A2, the labeling method is: the response time constant is set as: based on the load change response feature vectors to be trained, record the first time node when the load mutation is detected, continuously obtain the power values after the first time node until the power value tends to be stable, record the current time node as the second time node, and record the difference between the second time node and the first time node as the response time; the target power adjustment amount is set as the difference between the power value corresponding to the second time node and the power value of the first time node; the charge and discharge rate matching of the buffer device is set as the charge and discharge rate value of the corresponding buffer device within the response time.

[0022] Preferably, the preset progressive adjustment algorithm includes:

[0023] B1. Divide the response time into equal time intervals, and allocate the power adjustment amount for each time interval according to the preset S-shaped curve function to generate the scheduling instruction information of the power device, including the power adjustment amount sequence arranged in time order;

[0024] B2. Use the charge and discharge rate matching of the buffer device as the scheduling target when the collaborative power device executes the scheduling instruction information, that is, control the charge and discharge rate of the buffer device used to respond to the power device to be adjusted to the charge and discharge rate matching of the buffer device.

[0025] Preferably, S101 further includes: real-time monitoring of the electromagnetic radiation intensity information near the power equipment to obtain the change amount of the electromagnetic radiation intensity corresponding to the load change response feature vector when the power equipment meets the load mutation condition, which is used to reflect the degree of influence of the load mutation on the power environment;

[0026] The power equipment power adjustment behavior prediction model includes several prediction sub-models, and step A3 further includes:

[0027] S201, based on the change amount of the electromagnetic radiation intensity corresponding to all the to-be-trained load change response feature vectors, using the preset K-means clustering algorithm, dividing all the to-be-trained load change response feature vectors into several clusters, each cluster includes at least one to-be-trained load change response feature vector, and the center label of each cluster is the average value of all the electromagnetic radiation intensity change amounts in the cluster;

[0028] S202, based on several clusters, generating several sub-training sets, that is, each to-be-trained load change response feature vector in each cluster corresponds to a sub-training set;

[0029] S203, based on each sub-training set, training and learning the pre-selected neural network structure, continuously optimizing the model parameters, and generating the final sub-prediction model.

[0030] Preferably, S102 further includes:

[0031] Based on each power equipment, obtaining the change amount of the electromagnetic radiation intensity corresponding to its load change response feature vector, calculating the Euclidean distance values between the change amount of the electromagnetic radiation intensity and the center label of each cluster respectively, and inputting the load change response feature vector into the sub-prediction model corresponding to the cluster with the smallest Euclidean distance value to output the adjustment behavior information of the power equipment.

[0032] Preferably, S101 further includes: real-time collecting the azimuth angle information of the power equipment through the azimuth angle collecting device pre-installed on the outer shell of the power equipment , which is used to reflect the orientation of the power equipment in the current power scene space;

[0033] Before S103, the method further includes:

[0034] S301, based on each power equipment, according to the current azimuth angle θ, querying the pre-constructed azimuth angle-response rate mapping table to obtain the azimuth angle correction coefficient of the power equipment ;

[0035] S302, incorporating the current azimuth angle correction coefficient of the power equipment into the predicted adjustment behavior information to update the adjustment behavior information;

[0036] After the B1, it further includes: adjusting the power adjustment amount sequence of the power equipment according to the azimuth correction coefficient to obtain an updated power adjustment amount sequence.

[0037] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0038] By real-time monitoring the power of the power equipment and obtaining the load change response feature vector, the sudden load change situation can be detected in time, providing an accurate basis for subsequent scheduling adjustment; the power equipment power adjustment behavior prediction model predicts the adjustment behavior information according to the feature vector, enabling the power equipment to make reasonable responses when the load suddenly changes, avoiding the problem of frequent start and stop of the equipment caused by traditional sudden scheduling, reducing the mechanical wear of the equipment, and improving the service life of the equipment and the stability of the power system; the progressive adjustment algorithm divides the response time into equal time intervals and distributes the power adjustment amount according to the S-shaped curve function, making the power adjustment of the power equipment more stable and reducing the impact of rapid load changes on the equipment stability and the safety of the surrounding environment; at the same time, taking the charge and discharge rate matching of the buffer equipment as the collaborative control target, ensuring that the buffer equipment can respond to the load changes of the power equipment in time, realizing the collaborative operation of the power equipment and the buffer equipment, and effectively avoiding the problem of instantaneous voltage fluctuation caused by the mismatch of the charge and discharge rates of the buffer equipment.

[0039] By considering the change amount of the electromagnetic radiation intensity near the power equipment, aiming at the relationship between the load change of the power equipment and the radiation characteristics, incorporating the characteristics of different load change scenarios into the model training process, using the K-means clustering algorithm to divide the to-be-trained load change response feature vectors into several clusters, each cluster corresponding to different scenario characteristics, and then generating sub-prediction models for each cluster, enabling each sub-prediction model to be more focused on the prediction of specific scenario characteristics, greatly improving the accuracy of the prediction results compared with the unified prediction model. BRIEF DESCRIPTION OF THE DRAWINGS

[0040] Figure 1 It is a schematic flowchart of the management method of the power equipment according to the embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0041] To facilitate the understanding of the present invention, the present application will be described more comprehensively below with reference to the relevant drawings; the drawings show the preferred embodiments of the present invention, however, the present invention can be implemented in many different forms and is not limited to the embodiments described herein; on the contrary, these embodiments are provided to make the disclosure of the present invention more thorough and comprehensive.

[0042] It should be noted that the terms "vertical", "horizontal", "upper", "lower", "left", "right" and similar expressions used in this article are for illustrative purposes only and do not represent the only implementation.

[0043] Unless otherwise defined, all technical and scientific terms used in this article have the same meaning as those commonly understood by those skilled in the technical field to which this invention belongs; the terms used in the description of this invention in this article are only for the purpose of describing specific implementations and are not intended to limit this invention; the term "and / or" used in this article includes any and all combinations of one or more related listed items.

[0044] Embodiment 1: Figure 1 It is a schematic flowchart of the management method of the power equipment in the embodiment of this invention.

[0045] As Figure 1 shown, a management method of power equipment includes the following steps:

[0046] S101, monitor the operating parameters of each power equipment in the power scenario in real time, and obtain the load change response feature vector corresponding to the power equipment when the load mutation condition is met.

[0047] Specifically, the load mutation condition is set as: the load change amount of the power equipment is greater than the preset load change threshold corresponding to the equipment type to which the power equipment belongs; among them, different equipment types correspond to different load change thresholds, which are set according to the actual situation and expert experience and are used to reflect the mutation degree of the load change of the power equipment. When a certain mutation degree is reached, it indicates that the working condition of the power equipment has changed and subsequent dispatching adjustment of the power equipment is required.

[0048] Specifically, the load change response feature vector is set as , where s is the coding value of the equipment type to which the power equipment belongs, P is the current power value of the power equipment, is the load step change amount (the difference in power values before and after the load mutation when the load mutation is detected), is the load change rate (the load change rate is calculated by the method of numerical differentiation. For example, it is set as ); R is the response feature, which is set as , ( )], T is the current temperature of the power equipment, is the maximum power adjustment rate, ( ) is the efficiency decay coefficient; the acquisition method of the maximum power regulation rate is as follows: perform simulation operation on the power equipment in advance to obtain it. Conduct a power regulation experiment simulation on the power equipment, gradually increase or decrease the power regulation command within the allowable range of the power equipment, and record the actual power regulation rate of the power equipment at the same time. Through multiple experiments, confirm the maximum power regulation rate that the equipment can reach (record the power change curve, calculate the slope value, and take the maximum value of multiple tests) as the maximum power regulation rate characteristic of the power equipment; the efficiency decay coefficient is set as: , is the rated power value preset for the power equipment.

[0049] S102. Based on each power equipment, input the load change response feature vector into the pre-trained power equipment power regulation behavior prediction model, and output the corresponding regulation behavior information of the power equipment, including the target power regulation amount, response time, and charge and discharge rate matching of the buffer equipment.

[0050] In some embodiments, the acquisition method of the pre-trained power equipment power regulation behavior prediction model is as follows:

[0051] A1. Collect a large number of historical load change response feature vectors of power equipment that meet the corresponding load mutation conditions in history, and perform a preset screening mechanism to obtain the load change response feature vectors to be trained.

[0052] Among them, the preset screening mechanism is set as:

[0053] C1. Based on each historical load change response feature vector, calculate the power buffer matching value according to the following formula:

[0054]

[0055] Among them, K is the power buffer matching value, is the maximum voltage value absorbed or released by the buffer equipment responding to the power equipment within the response time, is the voltage difference of the corresponding power equipment within the response time (the difference between the maximum value and the minimum value of the voltage). The power buffer matching value reflects the degree to which the power buffer equipment can respond to the load mutation of the power equipment. If the maximum capacity of the power buffer equipment is large enough to completely absorb or release the power of the load mutation, the power matching value is 1; if the maximum capacity of the power buffer equipment is less than the power of the load mutation, the power matching value is less than 1, indicating that the buffer equipment cannot fully respond to the load mutation.

[0056] C2. Determine all historical load change response feature vectors with power buffer matching values greater than the preset matching threshold as the feature vectors of load change response to be trained. The preset matching threshold is set according to the actual situation and expert experience. For example, it is set to be slightly less than 1.

[0057] A2. Perform label annotation on the feature vectors of load change response to be trained, and set the annotation content as adjustment behavior information, including target power adjustment amount, response time, and charge-discharge rate matching of buffer devices.

[0058] Among them, the response time constant represents the response speed of the power device to load changes. It is a time constant used to describe the delay characteristics of the power device's response, reflecting the time characteristics required for the power device to transition from one stable state to another. The response time is set as follows: Based on the feature vectors of load change response to be trained, record the first time node when a load mutation is detected, continuously obtain the power values after the first time node until the power value tends to be stable (the determination condition for stability can be set that the slope value of the power change curve within a preset window is less than a preset threshold. The preset threshold is set according to expert experience and is used to measure the degree of power change. The preset window can be set to 1 s. Analyze the slope values of each preset window step by step until the slope value is less than the preset threshold for the first time, and record the time node corresponding to the first power value within this preset window as the stable time node), record the current time node as the second time node, and record the difference between the second time node and the first time node as the response time; the target power adjustment amount is set as the difference between the power value corresponding to the second time node and the power value of the first time node; the charge-discharge rate matching of the buffer device is set as the charge-discharge rate value of the corresponding buffer device within the response time.

[0059] A3. Use all the labeled feature vectors of load change response as the training set, train and learn the pre-selected neural network structure, and continuously optimize the model parameters to generate the final prediction model for the power adjustment behavior of the power device.

[0060] S103. Based on the adjustment behavior information, use the pre-set progressive adjustment algorithm to generate the scheduling instruction information of the power device.

[0061] In some embodiments, the pre-set progressive adjustment algorithm includes:

[0062] B1. Divide the response time into equal time intervals, and allocate the power adjustment amount for each time interval according to the preset S-shaped curve function to generate the scheduling instruction information of the power device, including a sequence of power adjustment amounts arranged in chronological order.

[0063] Specifically, step B1 includes:

[0064] Divide the response time T into N equal time intervals, and the number of intervals is dynamically adjusted according to the device type. For example, N = T / , is a preset reference time interval, which is set according to expert experience; the preset S-shaped curve function is:

[0065]

[0066] where, is the power adjustment amount for the i-th time interval, is the target power adjustment amount, k is a preset smoothing coefficient, which is set according to the response characteristics of the power equipment and expert experience, and determines the steepness of the curve. The larger the k value, the faster the power adjustment changes, and vice versa, the slower; is the start time of the i-th time interval, and T is the duration of the response time. The S-shaped curve distributes the power adjustment amount, which can make the power adjustment change more gently at the beginning and end, reducing the impact on the equipment. For example, for a large industrial motor, if the power suddenly changes greatly, it may cause excessive stress on components such as the motor bearings and windings, while the S-shaped curve adjustment can avoid this situation; many power equipment have a certain response delay and inertia and cannot reach the target power instantly. The S-shaped curve distributes the power adjustment amount, which can better conform to the actual response characteristics of the equipment and enable the equipment to transition to the target power state more naturally.

[0067] B2. Take the charge and discharge rate matching of the buffer device as the scheduling target when the collaborative power equipment executes the scheduling instruction information, that is, control the charge and discharge rate of the buffer device used to respond to the power equipment to be adjusted to the buffer device matching charge and discharge rate.

[0068] S104. Manage and control the power equipment according to the scheduling instruction information of the power equipment and the buffer device, realize the reasonable scheduling of the power equipment, and ensure power stability and load balance.

[0069] As an example, a certain industrial park has deployed a photovoltaic power generation system, a energy storage battery pack (as a power buffer device), and multiple industrial load devices. At this time, it is monitored that the power of a certain industrial load device suddenly increases. The encoded value of the device type to which this industrial load device belongs can be recorded as "001". The load change amount exceeds the preset load change threshold corresponding to this device type, and a load change response feature vector of this power device is generated: [001, 100kW, 30kW, 5kW / s, 50°C, 10kW / s, 0.95]. This load change response feature vector is input into a pre-trained power device power adjustment behavior prediction model for learning, and the target power adjustment amount of this power device is output as -20kW (reducing power), the response time is 5s, and the charge and discharge rate matching of the buffer device is 8kW / s (charging). Using the progressive adjustment algorithm, the 5s response time is divided into 5 equal-length time intervals of 1s each. According to the preset S-shaped curve function, a power adjustment amount sequence is generated: [-2, -5, -8, -10, -5]. Based on the generated scheduling instruction information, the industrial load device is controlled to gradually reduce power according to the power adjustment amount sequence, and at the same time, the energy storage battery pack is controlled to charge at a rate of 8kW / s. Thus, in this way, the coordinated operation of the power device and the buffer device is achieved, avoiding problems caused by frequent start-stop of equipment and rapid load changes, and ensuring the stability of power supply and load balance in the industrial park.

[0070] The technical solutions in the above embodiments of the present application at least have the following technical effects or advantages:

[0071] By real-time monitoring the power of the power device and obtaining the load change response feature vector, it is possible to timely detect the load mutation situation and provide an accurate basis for subsequent scheduling adjustment; the power device power adjustment behavior prediction model predicts the adjustment behavior information according to the feature vector, enabling the power device to make a reasonable response when the load mutates, avoiding the problem of frequent start-stop of equipment caused by traditional mutation-based scheduling, reducing the mechanical wear of the equipment, and improving the service life of the equipment and the stability of the power system; the progressive adjustment algorithm divides the response time into equal-length time intervals and distributes the power adjustment amount according to the S-shaped curve function, making the power adjustment of the power device more stable and reducing the impact of rapid load changes on the equipment stability and the safety of the surrounding environment; at the same time, taking the charge and discharge rate matching of the buffer device as the coordinated control target to ensure that the buffer device can timely respond to the load change of the power device, realizing the coordinated operation of the power device and the buffer device, and effectively avoiding the problem of instantaneous voltage fluctuation caused by the mismatch of the charge and discharge rate of the buffer device.

[0072] Combining the pre-trained power equipment power regulation behavior prediction model with the progressive adjustment algorithm can quickly and accurately generate scheduling instruction information for power equipment and buffer equipment. These instruction information have a clear time sequence and power regulation amount, which are convenient for actual control operations and improve the accuracy and executability of scheduling instructions.

[0073] Embodiment 2: In Embodiment 1, although the power equipment is managed through real-time monitoring and power regulation behavior prediction, a unified prediction model is used to handle the load changes or mutation scenarios of a large number of power equipment. Due to different load change scenarios having different characteristics, such as the amplitude, frequency of load changes, and the accompanying radiation change conditions, etc., the traditional technology may ignore the impact of load changes on the electromagnetic environment. It is difficult for a unified model to comprehensively and accurately capture these differences, resulting in inaccurate prediction results, which in turn affects the rationality of scheduling instructions.

[0074] For different load change scenarios, such as different amplitudes and frequencies of load changes, there are often different degrees of electromagnetic radiation changes. Incorporating the change amount of electromagnetic radiation intensity into consideration makes the description of each load change scenario no longer limited to traditional parameters such as load change amount and change rate, but adds a dimension related to the electromagnetic environment, thus more comprehensively and meticulously reflecting the essential characteristics of the load change scenario.

[0075] Therefore, the embodiments of the present application are optimized on the basis of the above embodiments.

[0076] In some embodiments, step S101 further includes: real-time monitoring the electromagnetic radiation intensity information near the power equipment, obtaining the change amount of electromagnetic radiation intensity corresponding to the load change response feature vector when the power equipment meets the load mutation condition, which is used to reflect the degree of influence of the load mutation on the power environment. It should be noted that how to obtain the electromagnetic radiation intensity can refer to relevant existing technologies, and the present invention will not elaborate on this.

[0077] Specifically, the power equipment power regulation behavior prediction model includes several prediction sub-models, and step A3 further includes:

[0078] S201, based on the change amount of electromagnetic radiation intensity corresponding to all the load change response feature vectors to be trained, using the preset K-means clustering algorithm, dividing all the load change response feature vectors to be trained into several clusters, each cluster includes at least one load change response feature vector to be trained, and the central label of each cluster is the average value of all the electromagnetic radiation intensity change amounts in the cluster.

[0079] Among them, the specific implementation process of the K-means clustering algorithm can refer to relevant existing technologies, and the present invention will not elaborate on this.

[0080] S202. Generate a number of sub-training sets based on several clusters, that is, each sub-training set corresponds to the load change response feature vectors to be trained in each cluster.

[0081] S203. Based on each sub-training set, train and learn the pre-selected neural network structure, continuously optimize the model parameters, and generate the final sub-prediction model.

[0082] In some embodiments, step S102 further includes:

[0083] Based on each power device, obtain the change amount of the electromagnetic radiation intensity corresponding to its load change response feature vector, calculate the Euclidean distance values between the change amount of the electromagnetic radiation intensity and each cluster center label respectively, input the load change response feature vector into the sub-prediction model corresponding to the cluster with the smallest Euclidean distance value, and output the adjustment behavior information of the power device.

[0084] Thus, each sub-prediction model focuses on the prediction of specific scenario characteristics, enabling the model to learn a more accurate relationship between load changes and adjustment behaviors in different scenarios. For example, in some load scenarios with high electromagnetic radiation changes, power devices may require faster or larger amplitude power adjustments, and the sub-prediction model can capture this specific relationship, thereby improving the adaptability and prediction accuracy of the model to different scenarios.

[0085] The technical solutions in the above embodiments of the present application at least have the following technical effects or advantages:

[0086] By considering the change amount of the electromagnetic radiation intensity near the power device, aiming at the relationship between the load change of the power device and the radiation characteristics, incorporating the characteristics of different load change scenarios into the model training process, using the K-means clustering algorithm to divide the load change response feature vectors to be trained into several clusters, each cluster corresponding to different scenario characteristics, and then generating a sub-prediction model for each cluster, enabling each sub-prediction model to focus more on the prediction of specific scenario characteristics. Compared with the unified prediction model, the accuracy of the prediction results is greatly improved;

[0087] According to the Euclidean distance between the change amount of the electromagnetic radiation intensity corresponding to the load change response feature vector of the power device and the cluster center label, select the most suitable sub-prediction model for prediction to ensure that the output adjustment behavior information better conforms to the actual load change scenario of the power device, which helps the power device make a more reasonable response during load mutations and avoids the problem of unreasonable scheduling caused by inaccurate prediction;

[0088] Accurate prediction and reasonable scheduling instructions can improve the stability, reliability, and load balance of the power system. In actual power scenarios, different load change scenarios may cause power equipment to face different challenges. Being able to better address these challenges ensures the stable operation of power equipment in various scenarios.

[0089] Embodiment 3: In actual power scenarios, the installation orientation of power equipment may vary due to factors such as site layout and wiring. For example, in some industrial plants, power equipment may be close to walls, large equipment, or other heat sources. The thermal effects and electromagnetic interference on equipment at different azimuth angles in different working states are different. These factors will affect the power regulation ability and response rate of the equipment. In Embodiment 1, the heat dissipation conditions and electromagnetic environment of power equipment at different azimuth angles may be different. Traditional methods do not consider these factors, resulting in inaccurate prediction and scheduling.

[0090] Therefore, the embodiments of the present application are optimized based on the above embodiments.

[0091] In some embodiments, step S101 further includes: real-time collecting the azimuth angle information of the power equipment through an azimuth angle collecting device (which can be set as an electronic compass or a preset position code) pre-installed on the outer shell of the power equipment , which is used to reflect the current orientation of the power equipment in the power scenario space.

[0092] In some embodiments, before step S103, the method further includes:

[0093] S301, based on each power equipment, query the pre-constructed azimuth angle-response rate mapping table according to the current azimuth angle θ to obtain the azimuth angle correction coefficient of the power equipment .

[0094] Specifically, the pre-constructed azimuth angle-response rate mapping table includes power equipment type, azimuth angle, and azimuth angle correction coefficient; the construction method is as follows:

[0095] D1. In the laboratory or actual power scenarios, build an experimental environment that can simulate different azimuth angles for each type of power equipment. A rotating platform can be used to change the azimuth angle of the equipment to ensure that the operating parameters of the equipment at each azimuth angle can be accurately measured;

[0096] D2. Apply the same load change to the power equipment at different azimuth angles (0° - 360°, at a certain angular interval, such as 10°), record the response rate of the equipment. The response rate can be determined by measuring parameters such as the change rate of the output power of the power equipment and the change rate of the current. To eliminate the influence of the numerical difference in the response rate between different equipment types, normalize the data. The min-max normalization method can be used to map the data to the interval [0, 1].

[0097] D3. Use a preset mathematical model (such as polynomial fitting, sine function fitting, etc.) to fit the normalized response rate data at different azimuth angles to generate the functional relationship between the response rate and the azimuth angle.

[0098] For example, for some types of power equipment, the response rate may change periodically with the azimuth angle. In this case, a sine function can be used for fitting: y = Asin(ωθ + φ) + B, where y is the normalized response rate, θ is the azimuth angle, and A, ω, φ, B are fitting parameters.

[0099] It should be noted that the accuracy and reliability of the fitting model can be verified by methods such as cross-validation. Divide the data set into a training set and a test set, use the training set for model fitting, and then use the test set to evaluate the prediction performance of the model. If the prediction error of the model is within an acceptable range, the fitting relationship is considered valid. Refer to the relevant prior art, and the present invention will not elaborate on this.

[0100] D4. Select the response rate at a reference azimuth angle (such as 0°, which can be specifically set by an expert based on the actual power scenario and the type of power equipment) as the reference response rate , and calculate the correction coefficient at each azimuth angle according to the relationship between the response rate and the azimuth angle obtained by fitting:

[0101]

[0102] where, is the response rate at azimuth angle θ, is the correction coefficient corresponding to azimuth angle θ, is the preset reference azimuth angle.

[0103] D5. Discretize the azimuth angle at a preset interval (such as 10°), calculate the correction coefficient corresponding to each discrete azimuth angle, and form an azimuth angle - correction coefficient mapping table. The mapping table can be represented in the form of a two-dimensional array or a dictionary. Generate an independent mapping table for each type of equipment to reflect the response characteristics of different equipment at different azimuth angles.

[0104] S302. The current azimuth angle correction coefficient of the power equipment Incorporate it into the predicted adjustment behavior information to update the adjustment behavior information.

[0105] In some embodiments, after step B1, the method further includes:

[0106] According to the azimuth correction coefficient, adjust the power adjustment amount sequence of the power equipment to obtain an updated power adjustment amount sequence. The power adjustment amount at the i-th time interval after the update is: 。

[0107] In summary, during the actual operation of power equipment, there are differences in factors such as its heat dissipation conditions and electromagnetic environment at different azimuth angles. For example, when the side of the power equipment faces the main circuit (θ = 90°), its maximum power adjustment rate may decrease due to limited heat dissipation conditions. If the influence of the azimuth angle on power adjustment is not considered and the scheduling is carried out according to the original power adjustment amount sequence, it may cause the equipment to be overloaded, affecting the service life of the equipment and the stability of the power system. Through azimuth response compensation, by adjusting the power adjustment amount according to the azimuth angle, although the total power adjustment amount has changed, it can better adapt to the actual operation conditions of the equipment at different azimuth angles, avoiding equipment failures caused by heat dissipation and other problems. In the long run, it is beneficial to the safe and stable operation of the power system. In actual scenarios, the operating environment of power equipment is complex and changeable, and the azimuth angle is only one of the influencing factors. For example, in some industrial plants, power equipment may be close to high-temperature areas or there may be other heat sources, and the thermal effects on the equipment at different azimuth angles are different. By adjusting the power adjustment amount through azimuth response compensation, the equipment can operate more reasonably in a complex environment, reducing the degradation of equipment performance and the occurrence of failures caused by environmental factors. Through azimuth correction, the prediction model originally trained based on historical data can better adapt to the operating conditions at different azimuth angles, enhancing the generalization ability and adaptability of the model. Even when encountering azimuth angle situations that have not appeared in historical data, the model can reasonably adjust the predicted value according to the azimuth correction coefficient to ensure that the prediction result is more in line with the actual situation.

[0108] The technical solutions in the embodiments of the present application above have at least the following technical effects or advantages:

[0109] Considering the influence of the azimuth angle on the response rate of power equipment, adjusting the power adjustment amount sequence through the azimuth correction coefficient makes the scheduling instruction more in line with the actual operation conditions of the equipment at different azimuth angles, improving the accuracy of prediction and scheduling. It can adapt to the differences in factors such as heat dissipation conditions and electromagnetic environment of power equipment at different azimuth angles, avoiding problems such as equipment overload and failure caused by azimuth angle factors, and enhancing the adaptability of the equipment in a complex environment and the stability of the power system.

[0110] The above are only the preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, various modifications and variations can be made to the present invention. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included within the protection scope of the present invention.

Claims

1. A management method for power equipment, characterized in that, Including: S101, monitoring the power of each power device in the power scenario in real time, and obtaining the load change response feature vector corresponding to the power device when the load mutation condition is met; S102, based on each power device, inputting the load change response feature vector into the pre-trained power device power adjustment behavior prediction model, and outputting the adjustment behavior information corresponding to the power device, including the target power adjustment amount, response time, and charge-discharge rate matching of the buffer device; the acquisition method of the power device power adjustment behavior prediction model is: A1. Collect historical load change response eigenvectors of a large number of power equipment in history that meet the corresponding load mutation conditions, and use a screening mechanism to obtain the eigenvectors of load change responses to be trained. The screening mechanism is as follows: Based on each historical load change response eigenvector, obtain the power buffer matching value according to the following formula: , where K is the power buffer matching value, is the maximum voltage value absorbed or released by the buffer device responding to the power equipment within the response time, is the voltage difference of the corresponding power equipment within the response time; determine all historical load change response eigenvectors with power buffer matching values greater than the preset matching threshold as the eigenvectors of load change responses to be trained; A2, performing label annotation on the to-be-trained load change response feature vector, and setting the annotation content as the adjustment behavior information; A3, using all the labeled load change response feature vectors as the training set, training and learning the pre-selected neural network structure, and continuously optimizing the model parameters to generate the final power device power adjustment behavior prediction model; S103, based on the adjustment behavior information, using the pre-set progressive adjustment algorithm to generate the scheduling instruction information of the power device, and realizing the reasonable scheduling of the power device; the progressive adjustment algorithm includes: B1, dividing the response time into equal time intervals, and distributing the power adjustment amount of each time interval according to the preset S-shaped curve function to generate the scheduling instruction information of the power device, including the power adjustment amount sequence arranged in chronological order; B2, using the charge-discharge rate matching of the buffer device as the scheduling target when the cooperative power device executes the scheduling instruction information, that is, controlling the charge-discharge rate of the buffer device used to respond to the power device to be adjusted to the charge-discharge rate matching of the buffer device.

2. The management method of the power equipment according to claim 1, characterized in that, The load mutation condition is set as: the load change amount of the power device is greater than the preset load change threshold corresponding to the device type to which the power device belongs; among them, different device types correspond to different load change thresholds, which are set according to the actual situation and expert experience and are used to reflect the mutation degree of the load change of the power device.

3. The management method of the power equipment according to claim 1, characterized in that, The load change response feature vector is set as , where s is the coding value of the device type to which the power device belongs, P is the current power value of the power device, is the load step change amount, is the load change rate; R is the response feature, set as , ( )], where T is the current temperature of the power device, is the maximum power adjustment rate, ( ) is the efficiency decay coefficient.

4. The management method of the power equipment according to claim 3, characterized in that, In A2, the annotation method is: the response time constant is set as: based on the to-be-trained load change response feature vector, record the first time node when the load mutation is detected, continuously obtain the power value after the first time node until the power value tends to be stable, record the current time node as the second time node, and record the difference between the second time node and the first time node as the response time; the target power adjustment amount is set as the difference between the power value corresponding to the second time node and the power value of the first time node; the charge-discharge rate matching of the buffer device is set as the charge-discharge rate value of the corresponding buffer device within the response time.

5. The management method of the power equipment according to claim 3, characterized in that, S101 further includes: monitoring the electromagnetic radiation intensity information near the power device in real time, and obtaining the electromagnetic radiation intensity change amount corresponding to the load change response feature vector of the power device when the load mutation condition is met, which is used to reflect the degree of influence of the load mutation on the power environment; The power device power adjustment behavior prediction model includes several prediction sub-models, and step A3 further includes: S201. Based on the electromagnetic radiation intensity change amounts corresponding to all the load change response feature vectors to be trained, use the preset K-means clustering algorithm to divide all the load change response feature vectors to be trained into several clusters. Each cluster includes at least one load change response feature vector to be trained, and the central label of each cluster is the average value of all the electromagnetic radiation intensity change amounts in this cluster; S202. Based on the several clusters, generate several sub-training sets, that is, the load change response feature vectors in each cluster correspond to a sub-training set; S203. Based on each sub-training set, perform training and learning on the pre-selected neural network structure, continuously optimize the model parameters, and generate the final sub-prediction model.

6. The management method of the power equipment according to claim 5, characterized in that, The S102 further includes: Based on each power device, obtain the electromagnetic radiation intensity change amount corresponding to its load change response feature vector, calculate the Euclidean distance values between the electromagnetic radiation intensity change amount and the central labels of each cluster respectively, and input the load change response feature vector into the sub-prediction model corresponding to the cluster with the smallest Euclidean distance value, and output the adjustment behavior information of this power device.

7. The management method of the power equipment according to claim 5, characterized in that, The S101 further includes: collecting the azimuth information of the power device in real time through an azimuth acquisition device pre-installed on the outer shell of the power device , which is used to reflect the current orientation of the power device in the power scene space; Before the S103, the method further includes: S301. Based on each power device, query the pre-constructed azimuth-response rate mapping table according to the current azimuth angle θ to obtain the azimuth correction coefficient of the power device ; S302. Incorporate the current azimuth correction factor of the power equipment into the predicted adjustment behavior information to update the adjustment behavior information; ​ After the B1, it further includes: According to the azimuth correction coefficient, adjust the power adjustment amount sequence of the power device to obtain the updated power adjustment amount sequence.

Images