A power optimization method and system based on planned curve tracking

Abstract

The present application relates to the field of power analysis, and more particularly to a power optimization method and system based on a planned curve tracking, the method first identifies the monitoring type of the target power network; then according to the monitoring type, the corresponding multi-dimensional standard template is called; then, the monitoring items in the multi-dimensional standard template are identified, and the multi-dimensional power data corresponding to the monitoring items are regularly collected; then the multi-dimensional power data is converted into to-be-processed data; then, according to the algorithm type label, the to-be-processed data is input into the corresponding time series analysis algorithm, and the multi-dimensional prediction curve of the target scheduling period is obtained by prediction; the template curve and the multi-dimensional prediction curve are compared and operated to obtain the comparison result; finally, the comparison result is input into the preset strategy matching library for information matching to obtain the correction instruction to drive the corresponding power execution device to act. Compared with the prior art, the method of the present application can obtain the power optimization solution set more quickly and reliably, and drive the corresponding power execution device to act.

Description

Technical Field

[0001] This invention relates to the field of power analysis technology. More specifically, this invention relates to a power optimization method and system based on planned curve tracking. Background Technology

[0002] In modern power systems, the controllability requirements of grid dispatch on both the generation and load sides are increasingly stringent. Traditional energy management systems typically employ "constant power" or "constant peak time" control strategies for energy dispatch. These approaches often focus only on active power, neglecting the multi-dimensional requirements for reactive power, power factor, and ramp rate during dispatch. Furthermore, these methods primarily rely on instantaneous responses based on current data, failing to consider frequent adjustments based on load / generation forecasts over a future period, which can easily lead to overshoot or undershoot, resulting in significant equipment losses.

[0003] To address the aforementioned issues, Chinese patent application CN121791341A proposes a multi-reactive-source static and dynamic optimization control method and system based on source-grid collaboration. First, based on ultra-short-term forecast data, it constructs and solves a unified optimization model integrating multiple types of reactive power sources from both the grid and power supply sides. This generates switching plans for discrete reactive power sources, reference curves for continuous reactive power sources and reactive power output from the power supply side, and the voltage optimization target range for key buses within a first preset time period. Then, using a digital twin model of the power grid, it performs ultra-real-time simulation of the optimization plan, simulating the system dynamics under a predefined fault set. If the simulation results violate transient voltage safety standards, the optimization model is triggered to be re-solved with transient safety as an additional constraint. Next, based on the optimization objectives and plans, and combined with high-refresh-rate real-time measurement data, it identifies system disturbances online. Depending on the severity of the disturbance, it adaptively selects and executes either a tracking adjustment mode or an emergency collaborative control mode. Finally, each reactive power source executes the final control command and uploads its execution status and operating boundary information for updating the forecast model, triggering rolling optimization, and conditional re-optimization.

[0004] However, according to the IEC 62116 standard in this field, power systems, as engineering application systems, place greater emphasis on system reliability and rapid response capabilities. Their response speed needs to be at least at the millisecond level. Furthermore, different power grid companies have significantly different data evaluation standards (such as curve smoothness and response time). Therefore, the technical solution of nesting multiple artificial intelligence models can easily fall into the computational power-time paradox, exacerbating the computational power black hole. The calculation results may fail to converge for a long time, resulting in no response or timeouts when actual power dispatching or power fault troubleshooting is needed. Therefore, existing power optimization methods incorporating artificial intelligence suffer from insufficient reliability and slow response speed. Summary of the Invention

[0005] To address the technical problems of unreliability and slow response speed in existing power optimization methods that incorporate artificial intelligence, this invention discloses a power optimization method and system based on planning curve tracking.

[0006] In a first aspect, the present invention discloses a power optimization method based on planned curve tracking, comprising:

[0007] S10: Identify the monitoring type of the target power network; S20: Retrieve the corresponding multidimensional standard template according to the monitoring type; the multidimensional standard template includes monitoring items, template curves, and algorithm type labels; S30: Identify monitoring items in the multidimensional standard template and periodically collect multidimensional power data corresponding to the monitoring items; S40: Convert multidimensional power data into data to be processed; S50: Based on the algorithm type label, input the data to be processed into the corresponding time series analysis algorithm to predict the multidimensional prediction curve of the target scheduling period; S60: Compare the template curve with the multidimensional prediction curve to obtain the comparison result; S70: Input the comparison results into the preset strategy matching library for information matching, and obtain the correction command to drive the corresponding power actuator to operate.

[0008] Beneficial Effects: This invention introduces a target power network and an adaptation mechanism. Based on the actual situation of the target power network, it selects a corresponding multi-dimensional standard template, extracts monitoring items from the template, and collects multi-dimensional power data corresponding to these items. This achieves adaptive monitoring based on template adjustment, directly eliminating a large amount of irrelevant data and improving overall data transmission efficiency. Furthermore, a matching time series analysis algorithm is invoked to fit and predict the data to be processed, obtaining a multi-dimensional prediction curve. Finally, a simple template comparison and matching mechanism is used to derive correction instructions to drive the corresponding power execution device. Unlike existing technologies, this invention's method for power optimization consistently produces a solution set in a shorter time.

[0009] Preferably, step S50 includes: The data to be processed is input into the corresponding time series analysis algorithm to calculate the predicted value of each time node within the target scheduling period; Based on the fundamental attributes of the predicted values, the predicted values ​​are divided into multiple groups of predicted time series data with different dimensions. The predicted time series data for each dimension are concatenated in chronological order to obtain the predicted curve for the corresponding dimension.

[0010] Preferably, the correction instruction includes a soft output instruction and a hard execution drive instruction; the power execution device includes a power disconnect switch, a circuit breaker, and an inverter; step S70 includes: If the correction instruction is a soft output instruction, retrieve the optimization report and control instruction set from the preset knowledge base and push them to the monitoring and interactive terminal; If the correction instruction is a hard-execution drive instruction, it drives the corresponding power disconnect switch, circuit breaker and / or inverter to perform the corresponding action.

[0011] Preferably, the multidimensional standard template also includes monitoring indicators, tracking frequency, target detection side, communication protocol and encryption standard.

[0012] Preferably, step S30 includes: Based on the tracking frequency and target detection side, multi-dimensional power data corresponding to the monitoring items are collected; among which, the target detection side includes the power source side, the power grid side, and the environment side.

[0013] Preferably, step S40 includes: The multidimensional power data is encrypted and converted according to communication protocols and encryption standards to obtain the data to be processed.

[0014] Preferably, before step S50, the method of the present invention further includes: Based on the algorithm type label, retrieve the corresponding time series analysis algorithm from the local terminal or the cloud; Data cleaning is performed on the decoded data to be processed.

[0015] Preferably, time series analysis algorithms include ARIMA algorithm, XGBoost algorithm, and LSTM neural network algorithm.

[0016] Preferably, the comparison operation method in step S60 includes point-to-point comparison and sliding window comparison.

[0017] Secondly, the present invention discloses a power optimization system based on planning curve tracking, used in the power optimization method based on planning curve tracking described in the first aspect. The system of the present invention includes: The data interface layer is used for: Identify the monitoring type of the target power network; retrieve the corresponding multidimensional standard template according to the monitoring type; identify the monitoring items in the multidimensional standard template, and periodically collect the multidimensional power data corresponding to the monitoring items; The data processing layer is used to convert multidimensional power data into data to be processed. The predictive analysis layer is used to input the data to be processed into the corresponding time series analysis algorithm according to the algorithm type label, and predict the multidimensional prediction curve of the target scheduling period. The comparison analysis layer is used to perform comparison operations between the template curve and the multidimensional prediction curve to obtain the comparison results; The execution control layer is used to input the comparison results into a preset strategy matching library for information matching, and obtain correction instructions to drive the corresponding power execution device to act.

[0018] The advantages of this invention, which differ from existing technologies, are as follows: (1) The method of the present invention can obtain the power optimization solution set more quickly and reliably, and drive the corresponding power actuator to operate.

[0019] (2) The method of the present invention has stronger adaptability and can be securely connected to the private security standards of different power grid units without modifying too many underlying configurations.

[0020] (3) The method of the present invention has certain prediction and optimization functions and can meet the assessment technical standards of multiple power units. Attached Figure Description

[0021] The above and other objects, features, and advantages of exemplary embodiments of the present invention will become readily apparent from the following detailed description taken in conjunction with the accompanying drawings. In the drawings, several embodiments of the invention are illustrated by way of example and not limitation, and like or corresponding reference numerals denote like or corresponding parts, wherein: Figure 1 This is a flowchart of the power optimization method based on planning curve tracking in Embodiment 1 of the present invention; Figure 2 This is a schematic diagram of the power optimization system based on planned curve tracking in Embodiment 2 of the present invention. Detailed Implementation

[0022] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some, not all, of the embodiments of the present invention. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0023] The specific embodiments of the present invention will now be described in detail with reference to the accompanying drawings.

[0024] Example 1 like Figure 1 As shown, this embodiment discloses a power optimization method based on planned curve tracking, including: S10: Identify the monitoring type of the target power network.

[0025] It should be explained that power networks can be classified into five categories according to voltage, from low to high: low-voltage grids, medium-voltage grids, high-voltage grids, ultra-high-voltage grids, and extra-high-voltage grids. They can also be classified by current characteristics into AC grids and DC grids. Furthermore, they can be classified by region or function into industrial grids, residential grids, commercial grids, or photovoltaic grid-connected grids. These classifications are merely illustrative and do not represent all types of power networks. Identification of power network types can be achieved directly using existing power network topology analysis methods or voltage level labeling. After analysis and identification, the target power network is assigned a corresponding classification identifier to characterize the monitoring type.

[0026] S20: Retrieve the corresponding multidimensional standard template based on the monitoring type.

[0027] In this embodiment, the multidimensional standard template includes monitoring items, template curves, algorithm type labels, monitoring indicators, tracking frequency, target detection side, communication protocol, and encryption standard. Monitoring items refer to basic power-related data of the target power network, such as energy storage on the power source side, SOC (State of Charge) on the power source side, available capacity on the grid side, grid point voltage on the grid side, grid frequency on the grid side, real-time switching power on the grid side, and environmental irradiance, wind speed, and temperature. It should be noted that if the target power network is not a photovoltaic or wind power grid, environmental power-related data can be ignored. Template curves are multidimensional curves pre-planned for the target power network according to the target scheduling cycle, and vary depending on the region or type of power network. Algorithm type labels indicate the corresponding algorithm used by the algorithm scheduler and can be considered as identifiers. Monitoring indicators are the monitoring standards corresponding to the monitoring items; for example, the real-time switching power must not exceed 5% of the corresponding time period of the template curve within a certain period. Tracking frequency can be considered a data sampling period, which can be 1 minute / time or 5 minutes / time, specifically configured according to the hardware storage performance of the target power network. The target detection side refers to the power supply side, power grid side, and environment side mentioned above. Communication protocols and encryption standards differ across different power units. This embodiment does not innovate communication protocols or encryption standards; it only supports adaptability access and standard acquisition.

[0028] S30: Identify monitoring items in the multidimensional standard template and periodically collect multidimensional power data corresponding to the monitoring items.

[0029] Furthermore, step S30 above includes: Based on the tracking frequency and target detection side, multidimensional power data corresponding to the monitoring items are collected.

[0030] To improve the adaptability and reliability of the method in the embodiments, step S30 above also needs to combine the tracking frequency and the target detection side to collect multi-dimensional power data corresponding to the monitoring items. For example, assuming the target power network is a power network formed by the grid connection of a photovoltaic DC grid and an AC high-voltage grid, the target detection sides that need to be monitored include the power supply side, the grid side, and the environmental side. In this case, for the power supply side, the real-time power, SOC, and available capacity of the photovoltaic DC grid need to be collected; for the grid side, the voltage, frequency, and real-time switching power at the grid connection point need to be collected; and for the environmental side, irradiance and temperature need to be collected. The collection of the above data mainly relies on smart meters, photovoltaic and temperature monitoring devices.

[0031] Through the above step S30, selective power data acquisition based on multidimensional standard templates can be achieved. Unlike the global data acquisition in the prior art, this greatly reduces the amount of data processing and allows for targeted data analysis and adjustment based on the actual situation, thereby improving the reliability and efficiency of this implementation method.

[0032] S40: Convert multidimensional power data into data to be processed.

[0033] In step S40 above, the multidimensional power data needs to be encrypted and converted according to the communication protocol and encryption standard to obtain the data to be processed.

[0034] The above step S40 can ensure energy information security and reliable information transmission. The method in this embodiment mainly provides a standard alignment interaction module to support the client in black-box configuration of communication protocols and encryption standards.

[0035] Following step S40, the corresponding time series analysis algorithm needs to be retrieved from the local terminal or cloud based on the algorithm type label, and the decoded data to be processed needs to be cleaned. The data cleaning process includes missing value interpolation, outlier removal, and data smoothing. This approach allows for the rapid acquisition of highly adaptable time series analysis algorithms to support subsequent data analysis; data cleaning of the decoded data strengthens data characteristics and mitigates the impact of outliers on the data fitting process.

[0036] S50: Based on the algorithm type label, input the data to be processed into the corresponding time series analysis algorithm to predict the multidimensional prediction curve of the target scheduling period.

[0037] In this embodiment, the time series analysis algorithms mainly include the ARIMA algorithm, the XGBoost algorithm, and the LSTM neural network algorithm. These algorithms can balance accuracy and timeliness requirements. The target scheduling period refers to the forward prediction period of the current time, which can be 10 minutes, 30 minutes, or 1 hour.

[0038] Furthermore, step S50 above specifically includes: First, the data to be processed is input into the corresponding time series analysis algorithm to calculate the predicted value of each time node within the target scheduling period. Then, based on the basic attributes of the predicted values, the predicted values ​​are divided into multiple groups of predicted time series data of different dimensions. Finally, the predicted time series data of each dimension are connected in chronological order to obtain the prediction curve of the corresponding dimension.

[0039] The above steps mainly involve the construction of the fitted prediction curve. Unlike existing technologies, the above scheme involves a multi-dimensional curve fitting process that includes the power supply side, the power grid side, and / or the environment side. In this process, the predicted values ​​need to be screened according to basic attributes (output unit / monitoring node), and then the classified data are fitted according to a predetermined queue or parallel mechanism to obtain fitted prediction curves in multiple dimensions.

[0040] S60: Compare the template curve with the multidimensional prediction curve to obtain the comparison result.

[0041] Specifically, the above comparison operation methods include point-to-point comparison and sliding window comparison.

[0042] If it is a point-to-point comparison, the algorithm for the measurement deviation is as follows:

[0043] In the formula, Indicates the degree of deviation. This represents the standard value of the data item in the corresponding dimension of the template curve at the corresponding time point; This represents the value of the corresponding dimension curve in the multidimensional prediction curve at the corresponding time point.

[0044] If a sliding window comparison is used, all data within the sliding window segment need to be averaged first, and then fed into the algorithm described above for calculation.

[0045] After obtaining the deviation of the multidimensional data items, a multidimensional comparison is performed in conjunction with the monitoring indicators, and the comparison results are output. For example, the comparison results are presented intuitively in a table below.

[0046]

[0047] S70: Input the comparison results into the preset strategy matching library for information matching, and obtain the correction command to drive the corresponding power actuator to operate.

[0048] Regarding the data matching process for the strategy matching library, the configuration topologies of power execution devices differ among different power units. First, the location or quantity of power execution devices can be directly determined using graph topology methods. Then, an information mapping matching table is constructed. For example, this information mapping matching table can be defined as:

[0049] It should be explained that, generally, the correction instruction should be a hard-execution drive instruction, which directly drives the corresponding relay to turn on / off through signal conversion, thereby triggering the corresponding power execution device to act in advance. However, in order to ensure that there is a final solution and to perform timely scheduling optimization when an anomaly is actually predicted, the above correction instruction should also include a soft output instruction to overcome the functional dead zone anomaly that is not recorded in the strategy matching library or cannot be matched, thereby further improving the reliability of the method in this embodiment. More specifically, the above step S70 includes: S71: If the correction instruction is a soft output instruction, retrieve the optimization report and control instruction set from the preset knowledge base and push them to the monitoring and interactive terminal.

[0050] Furthermore, in comparison results other than those under normal circumstances, if a specific hard-execution drive instruction cannot be matched from the strategy matching library, the comparison result is pushed to the monitoring interaction terminal. At the same time, the optimization report and control instruction set in the knowledge base are pushed to the monitoring interaction terminal. In this embodiment, the optimization report records the past operation records and precautions of the target power system, while the control instruction set records the power execution devices that support switching operations of the target power system and the corresponding control instructions. The push of the above-mentioned optimization report and control instruction set can assist online maintenance personnel in performing safe and compliant operations.

[0051] Furthermore, when online maintenance personnel operate based on the comparison results, they need to enter personal information for registration and verify that at least one person is present to prevent malicious individual behavior or unauthorized operations due to a lack of security awareness.

[0052] S72: If the correction instruction is a hard-execution drive instruction, it drives the corresponding power disconnect switch, circuit breaker and / or inverter to perform the corresponding action.

[0053] Through the above steps S71-S72, the method of this embodiment can achieve a solution output in extreme scenarios, ensure safe and compliant execution, avoid computing power black holes, and respond proactively when power scheduling or power fault troubleshooting is needed in practice.

[0054] Based on the above technical discussion, the method of this embodiment has at least the following advantages: (1) It breaks through the limitations of single-dimensional active power prediction and realizes the coordinated prediction and optimization of multiple indicators such as active power, reactive power, and unit combination, which can meet the technical standard requirements.

[0055] (2) It mainly adopts "adaptive template fitting + knowledge base matching" for multi-dimensional prediction and decision output, overcoming the computing power black hole or output unsolvable problem of nested multiple artificial intelligence models in the existing technology.

[0056] (3) It has stronger adaptive capabilities and a higher data processing rate. The prediction rate can reach the ms level, which will not cause excessive computing power burden on the computing unit to which the prediction analysis layer belongs, thereby ensuring the safe, reliable and efficient operation of the power monitoring and optimization system.

[0057] Example 2 like Figure 2 As shown, this embodiment discloses a power optimization system based on planning curve tracking, used in the power optimization method based on planning curve tracking described in Embodiment 1. The system of this invention includes: The data interface layer is used for: Identify the monitoring type of the target power network; retrieve the corresponding multidimensional standard template according to the monitoring type; identify the monitoring items in the multidimensional standard template, and periodically collect the multidimensional power data corresponding to the monitoring items; The data processing layer is used to convert multidimensional power data into data to be processed. The predictive analysis layer is used to input the data to be processed into the corresponding time series analysis algorithm according to the algorithm type label, and predict the multidimensional prediction curve of the target scheduling period. The comparison analysis layer is used to perform comparison operations between the template curve and the multidimensional prediction curve to obtain the comparison results; The execution control layer is used to input the comparison results into a preset strategy matching library for information matching, and obtain correction instructions to drive the corresponding power execution device to act.

[0058] In the description of this specification, "multiple" means at least two, such as two, three or more, etc., unless otherwise expressly and specifically defined.

[0059] While this specification has shown and described numerous embodiments of the invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Many modifications, alterations, and alternatives will occur to those skilled in the art without departing from the spirit and essence of the invention. It should be understood that various alternatives to the embodiments of the invention described herein may be employed in the practice of this invention.

Claims

1. A power optimization method based on planned curve tracking, characterized in that, include: S10: Identify the monitoring type of the target power network; S20: Retrieve the corresponding multidimensional standard template according to the monitoring type; wherein, the multidimensional standard template includes monitoring items, template curves, and algorithm type labels; S30: Identify the monitoring items in the multidimensional standard template and periodically collect the multidimensional power data corresponding to the monitoring items; S40: Convert the multidimensional power data into data to be processed; S50: Based on the algorithm type label, input the data to be processed into the corresponding time series analysis algorithm to predict the multidimensional prediction curve of the target scheduling period; S60: Compare the template curve with the multidimensional prediction curve to obtain the comparison result; S70: Input the comparison result into a preset strategy matching library for information matching to obtain a correction instruction to drive the corresponding power execution device to operate.

2. The power optimization method based on planned curve tracking according to claim 1, characterized in that, Step S50 includes: The data to be processed is input into the corresponding time series analysis algorithm to calculate the predicted value of each time node within the target scheduling period; Based on the basic attributes of the predicted values, the predicted values ​​are divided into multiple groups of predicted time series data with different dimensions. The predicted time series data for each dimension are concatenated in chronological order to obtain the predicted curve for the corresponding dimension.

3. The power optimization method based on planned curve tracking according to claim 1, characterized in that, The correction instructions include soft-output instructions and hard-execution drive instructions; The power actuator includes a power disconnect switch, a circuit breaker, and an inverter; Step S70 includes: If the correction instruction is a soft output instruction, the optimization report and control instruction set are retrieved from the preset knowledge base and pushed to the monitoring and interactive terminal; If the correction instruction is a hard-execution drive instruction, it drives the corresponding power disconnect switch, circuit breaker and / or inverter to perform the corresponding action.

4. The power optimization method based on planned curve tracking according to claim 1, characterized in that, The multidimensional standard template also includes monitoring indicators, tracking frequency, target detection side, communication protocol and encryption standard.

5. The power optimization method based on planning curve tracking according to claim 4, characterized in that, Step S30 includes: Based on the tracking frequency and the target detection side, multidimensional power data corresponding to the monitoring item are collected; wherein, the target detection side includes the power supply side, the power grid side, and the environment side.

6. The power optimization method based on planning curve tracking according to claim 4, characterized in that, Step S40 includes: The multidimensional power data is encrypted and converted according to the communication protocol and the encryption standard to obtain the data to be processed.

7. The power optimization method based on planning curve tracking according to claim 4, characterized in that, Prior to step S50, the method further includes: Based on the algorithm type label, retrieve the corresponding time series analysis algorithm from the local terminal or the cloud; Data cleaning is performed on the decoded data to be processed.

8. The power optimization method based on planned curve tracking according to claim 1, characterized in that, The time series analysis algorithms include ARIMA algorithm, XGBoost algorithm, and LSTM neural network algorithm.

9. The power optimization method based on planned curve tracking according to claim 1, characterized in that, The comparison operation methods in step S60 include point-to-point comparison and sliding window comparison.

10. A power optimization system based on planned curve tracking, characterized in that, The system for the power optimization method based on planned curve tracking as described in any one of claims 1-9, the system comprising: The data interface layer is used for: Identify the monitoring type of the target power network; retrieve the corresponding multidimensional standard template according to the monitoring type; identify the monitoring items in the multidimensional standard template, and periodically collect the multidimensional power data corresponding to the monitoring items; The data processing layer is used to convert the multidimensional power data into data to be processed; The predictive analysis layer is used to input the data to be processed into the corresponding time series analysis algorithm according to the algorithm type label, and predict the multidimensional prediction curve of the target scheduling period. The comparison analysis layer is used to perform comparison operations between the template curve and the multidimensional prediction curve to obtain the comparison results; The execution control layer is used to input the comparison results into a preset strategy matching library for information matching, and obtain correction instructions to drive the corresponding power execution device to operate.

Images