A Substation Air Conditioning Group Control Energy-Saving Method Based on Load Forecasting

Abstract

This invention relates to the field of substation air conditioning control technology, specifically a substation air conditioning group control energy-saving method based on load forecasting. The method includes: collecting and storing historical grid load and air conditioning historical operation data; based on the predicted grid load curve, filtering historical load periods with similar characteristics, extracting corresponding air conditioning operation records to form an analysis set, analyzing to obtain a set of control modes that meet temperature control constraints, and selecting the one with the highest adaptability to future loads as the basic control framework; identifying adjustable periods, coupling user control commands with future loads to generate a preliminary group control strategy, inputting a conflict verification model to verify conflicts, and outputting a report containing conflict periods. This method can achieve precise adaptation of air conditioning control to grid load, reduce ineffective energy consumption, avoid operational conflicts in advance, ensure stable operation of substation equipment, and improve the energy efficiency and reliability of air conditioning group control.

Description

Technical Field

[0001] This invention relates to the field of substation air conditioning control technology, and in particular to a substation air conditioning group control energy-saving method based on load forecasting. Background Technology

[0002] Substation air conditioning systems are crucial auxiliary systems for ensuring the stable operation of substation equipment. Their energy consumption accounts for a significant portion of total energy consumption, and their operational status directly impacts equipment heat dissipation and the overall energy efficiency of the substation. Existing substation air conditioning group control technologies often employ fixed temperature control thresholds or rely solely on simple start-stop adjustments based on real-time grid load. While some solutions collect historical air conditioning operating data, they fail to correlate this data with grid load forecasts. Control decisions depend on a single-dimensional parameter, making dynamic adaptation to grid load changes impossible.

[0003] Existing control schemes fail to consider future grid load trends and cannot optimize air conditioning control strategies based on load fluctuations, resulting in insufficient adaptability of air conditioning operation to grid load and ineffective energy consumption. Furthermore, after receiving user control commands, existing schemes do not verify the compatibility of the commands with substation temperature control constraints and grid load demands, easily leading to conflicts between control commands and actual operational needs, affecting equipment operational stability, and making it difficult to balance the energy-saving goals and operational reliability of group air conditioning control. Therefore, a substation air conditioning group control technology solution is needed that combines grid load forecasting to optimize air conditioning control mode selection and can perform conflict verification of control commands. Summary of the Invention

[0004] The purpose of this invention is to overcome the shortcomings of existing technologies and propose a substation air conditioning group control energy-saving method based on load forecasting.

[0005] To achieve the above objectives, the present invention adopts the following technical solution: a substation air conditioning group control energy-saving method based on load forecasting, comprising: Collect and store historical power grid load data and substation air conditioning historical operation data, wherein the air conditioning historical operation data includes multiple operation record units; Based on the future power grid load curve, several historical load periods with similar characteristics to the future power grid load are selected from the historical power grid load data; From the historical air conditioning operation data, extract the historical air conditioning operation record units corresponding to each historical load period to form a set of operation records to be analyzed; The set of operation records to be analyzed is analyzed to identify a set of air conditioning control modes that meet the temperature control constraints of the substation. The set of air conditioning control modes includes multiple control modes. The control mode with the highest adaptability to the future power grid load curve is selected from the set of air conditioning control modes as the basic control framework. Identify adjustable runtime segments within the basic control framework; The user's control commands for the controllable operating period are coupled with the predicted power grid load curve to form a coupling analysis result. The coupling analysis result is then integrated with the basic control framework to generate a preliminary air conditioning group control execution strategy. The preliminary air conditioning group control execution strategy is input into a pre-built conflict verification model to verify whether there are any instructions that conflict with temperature control constraints or power grid load requirements. The conflict verification model outputs a verification report containing the conflict period.

[0006] As a further aspect of the present invention, selecting multiple historical load periods from historical power grid load data that are similar to future power grid load characteristics includes the following steps: The load peak characteristics, load valley characteristics, load change rate characteristics, and load duration characteristics are extracted from the predicted future power grid load curve. The historical power grid load data is traversed, and historical load features are extracted using a fixed time window as the step size. A load feature vector is constructed, which is composed of the load peak feature, load valley feature, load change rate feature and load duration feature according to a preset weight; Calculate the distance between the load feature vector corresponding to the predicted future power grid load curve and the historical load feature vector corresponding to each historical window; A distance threshold is set, and all historical time windows with distance values ​​less than the distance threshold are identified as historical load periods similar to the future power grid load characteristics.

[0007] As a further aspect of the present invention, extracting the air conditioner historical operation record unit corresponding to each historical load period from the air conditioner historical operation data includes the following steps: A time index is created for each historical load period, and the time index includes a start timestamp and an end timestamp; In the historical operating data of the air conditioner, locate the time marker corresponding to the operating record unit; Determine whether the timestamp corresponding to the running record unit falls completely within the time interval defined by the start timestamp and the end timestamp; If it falls completely into the range, the operation record unit is determined to be a valid operation record unit corresponding to the historical load period. All valid operation record units corresponding to all historical load periods are aggregated to form the set of operation records to be analyzed.

[0008] As a further aspect of the present invention, identifying the set of air conditioning control modes that satisfy the temperature control constraints of the substation includes the following steps: Each valid operation record unit in the set of operation records to be analyzed corresponds to a set of state sequences of air conditioning equipment. The temperature change sequence of the substation's internal environment was extracted from the effective operation record unit; The temperature change sequence of the substation's internal environment is compared with the preset substation temperature control constraint range. Valid operation record units whose temperature change sequence is entirely within the substation temperature control constraint range are selected and defined as compliant operation record units. Pattern clustering is performed on the state sequences of the air conditioning equipment corresponding to the compliant operation record units, and compliant operation record units with state sequence similarity higher than a preset threshold are grouped into the same category; Extract the most representative state sequence from each category to form an air conditioning control mode; All extracted air conditioning control modes constitute the air conditioning control mode set.

[0009] As a further aspect of the present invention, selecting the control mode with the highest adaptability to the future power grid load curve from the set of air conditioning control modes as the basic control framework includes the following steps: For each air conditioning control mode in the set of air conditioning control modes, predict the air conditioning power consumption curve that it will generate when it is running in the future time period corresponding to the predicted grid load curve. The predicted future power grid load curve is superimposed with each predicted air conditioning power consumption curve to form multiple predicted total load curves. Based on the power grid smooth operation index, the smoothness score of each predicted total load curve is calculated. From the set of air conditioning control modes, select the air conditioning control mode that gives the highest smoothness score to the corresponding predicted total load curve, and use its structure as the basic control framework.

[0010] As a further aspect of the present invention, identifying adjustable runtime segments in the basic control framework includes the following steps: In the predicted air conditioning power consumption curve corresponding to the basic control framework, the high power consumption period of the air conditioner that overlaps with the peak load period in the power grid load curve is identified. During the high power consumption period of the air conditioner, based on the redundancy adjustment capability of the air conditioning equipment, the sub-periods with the potential to reduce power without affecting the temperature control constraints are marked as the adjustable operating periods. In the basic control framework, the control instruction segments corresponding to the adjustable runtime segments are highlighted or specially marked.

[0011] As a further aspect of the present invention, the coupling analysis of the user's control commands for the controllable operating period with the predicted future power grid load curve includes the following steps: The control command includes the adjustment amount of the air conditioning equipment operating parameters within a specific controllable operating period. The control command is applied to the basic control framework to simulate and calculate the change in air conditioning power consumption and the corresponding change in temperature of the substation's internal environment after the adjustment. Determine whether the temperature change of the substation's internal environment after adjustment exceeds the substation's temperature control constraint range; if it does, correct the control command. In the predicted future power grid load curve, the future power grid load point corresponding to the specific controllable operating period is located. The change in air conditioning power consumption is time-series aligned with the future power grid load point. The impact of the change in air conditioning power consumption on the power grid load fluctuation during the time period of the future power grid load point is analyzed, and a coupled analysis result containing quantitative indicators of the impact is generated.

[0012] As a further aspect of the present invention, the conflict verification model is constructed based on a rule-based reasoning engine; The conflict verification model has a set of pre-set verification rules, which include temperature control rules, load fluctuation constraint rules, and equipment start-up and shutdown safety interval rules. The conflict verification model compares each control command in the preliminary air conditioning group control execution strategy with all verification rules one by one. When a control command triggers any verification rule, the time period corresponding to the control command is marked as a conflict time period, and the specific verification rule information triggered is recorded. The conflict verification model outputs a verification report containing information on all conflict periods and the specific verification rules that triggered them.

[0013] As a further aspect of the present invention, after generating the initial air conditioning group control execution strategy, a strategy optimization step is also included: Based on the impact quantification index in the coupling analysis results, the adjustable running segments during non-conflict periods in the preliminary air conditioning group control execution strategy are prioritized. Based on the verification report, the control commands that are in conflict periods are corrected according to the verification rules they trigger. By combining the priority ranking results and instruction correction results, an optimization suggestion list is generated, and the optimization suggestion list is merged with the preliminary air conditioning group control execution strategy to form the final air conditioning group control energy-saving operation strategy.

[0014] As a further aspect of the present invention, the execution of the final air conditioning group control energy-saving operation strategy includes the following steps: The final air conditioning group control energy-saving operation strategy is decomposed into a discrete set of time-sequenced control instructions; Establish a communication link between the timing control instruction set and the control terminal of the substation air conditioning system, and issue the instructions in chronological order; During the execution process, the actual temperature data of the substation and the actual load data of the power grid are collected in real time. The actual temperature data is compared with the preset temperature control constraint range of the substation, and the actual load data is compared with the predicted power grid load curve. When the actual temperature data or actual load data is detected to deviate from the preset deviation threshold, the strategy regeneration process is triggered to generate a new energy-saving operation strategy for the air conditioning group control, starting from the current moment.

[0015] Compared with the prior art, the advantages and positive effects of the present invention are as follows: Based on the predicted power grid load curve, historical load periods with similar characteristics to the future power grid load are selected from historical power grid load data. The corresponding historical air conditioning operation records for these load periods are extracted and compiled into a set of operation records to be analyzed. Through analysis of this set, a set of air conditioning control modes that meet the substation temperature control constraints is identified. Then, the control mode with the highest adaptability to the future power grid load curve is selected from this set as the basic control framework. This scheme allows the air conditioning control mode to adapt to future power grid load changes in advance, avoiding control lag problems caused by a lack of load prediction in conventional control, reducing ineffective energy consumption caused by mismatch between control mode and load, and ensuring that air conditioning operation always meets the substation temperature control requirements, achieving a synergy between energy saving and equipment heat dissipation needs.

[0016] Within the selected basic control framework, adjustable operating periods are identified. User-issued control commands for these adjustable operating periods are coupled with predicted grid load curves for future analysis. Based on the coupling analysis results, a preliminary air conditioning group control execution strategy is generated by integrating it with the basic control framework. This preliminary strategy is then input into a pre-built conflict verification model to verify whether there are any commands in the strategy that conflict with substation temperature control constraints or grid load demands. The conflict verification model directly outputs a verification report including the conflict periods. This scheme can identify conflicts between control commands and temperature control / grid load demands in advance, avoiding equipment malfunctions caused by blindly executing control commands in conventional control, reducing ineffective control operations, improving the operational stability of the air conditioning group control system, further optimizing energy consumption allocation, and achieving more precise energy-saving control. Attached Figure Description

[0017] Figure 1 This is a flowchart of a substation air conditioning group control energy-saving method based on load forecasting, as described in this invention. Figure 2A flowchart for filtering historical load periods from historical power grid load data that are similar to future power grid load characteristics; Figure 3 A flowchart for identifying the set of air conditioning control modes that meet the temperature control constraints of the substation; Figure 4 This section defines the basic power consumption curve and adjustable time period for the air conditioner. Figure 5 A comparison of historical and projected power grid load curves. Detailed Implementation

[0018] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0019] In the description of this invention, it should be understood that the terms "length," "width," "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "top," "bottom," "inner," and "outer," etc., indicating orientation or positional relationships, are based on the orientation or positional relationships shown in the accompanying drawings and are only for the convenience of describing the invention and simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of the invention. Furthermore, in the description of this invention, "a plurality of" means two or more, unless otherwise explicitly specified.

[0020] See Figure 1 This invention provides a substation air conditioning group control energy-saving method based on load forecasting, the specific method including: Historical power grid load data and substation air conditioning historical operation data containing multiple operation record units are collected and stored. Based on the predicted power grid load curve, several historical load periods similar to the characteristics of the future power grid load are selected from the historical power grid load data. Historical operation record units corresponding to each historical load period are extracted from the air conditioning historical operation data, forming a set of operation records to be analyzed. This set of operation records is analyzed to identify a set of air conditioning control modes that meet the substation temperature control constraints and include multiple control modes. The control mode with the highest adaptability to the future power grid load curve is selected from the set of air conditioning control modes as the basic control framework. Adjustable operating periods are identified within the basic control framework. User control commands for adjustable operating periods are coupled with the predicted power grid load curve to form a coupling analysis result. This coupling analysis result is integrated with the basic control framework to generate a preliminary air conditioning group control execution strategy. The preliminary air conditioning group control execution strategy is input into a pre-built conflict verification model to verify whether there are any commands that conflict with temperature control constraints or power grid load demands. The conflict verification model outputs a verification report containing conflict periods.

[0021] In one embodiment of the present invention, multiple historical load periods similar to future grid load characteristics are selected from historical grid load data. The process includes the following steps. (See also...) Figure 2 The system extracts peak load characteristics, valley load characteristics, load change rate characteristics, and load duration characteristics from the predicted power grid load curve. It iterates through historical power grid load data, extracting historical load characteristics in fixed time window steps. A load feature vector is constructed, composed of peak load characteristics, valley load characteristics, load change rate characteristics, and load duration characteristics weighted according to preset weights. The distance between the load feature vector corresponding to the predicted power grid load curve and the historical load feature vector corresponding to each historical window is calculated. A distance threshold is set, and all historical time windows with distance values ​​less than the threshold are identified as historical load periods similar to the future power grid load characteristics.

[0022] In practical implementation, a substation air conditioning group control energy-saving method based on load forecasting involves screening multiple historical load periods from historical power grid load data that are similar to the characteristics of future power grid loads. The predicted power grid load curve may correspond to the full-day load curve for tomorrow, i.e., April 3rd, while the historical power grid load data includes load records for each day of the past year. Load peak characteristics, load valley characteristics, load change rate characteristics, and load duration characteristics are extracted from the predicted power grid load curve. The load peak characteristic can be the maximum power value in the curve, the load valley characteristic is the minimum power value, the load change rate characteristic is obtained by calculating the maximum absolute value of power change per unit time, and the load duration characteristic can be defined as the length of time during which the power remains above a certain level.

[0023] In some embodiments, historical power grid load data is traversed, and historical load features are extracted using a fixed time window as the step size. The fixed time window can be set to 24 hours, with a step size of 1 hour. This means that starting from the first hour of the historical data, a 24-hour historical window is extracted every hour, and the load peak characteristics, load valley characteristics, load change rate characteristics, and load duration characteristics of that window are calculated. A load feature vector is constructed, which is composed of load peak characteristics, load valley characteristics, load change rate characteristics, and load duration characteristics according to preset weights. The preset weights can be set according to power grid operation experience; for example, the weight of load peak characteristics is 0.4, the weight of load valley characteristics is 0.3, the weight of load change rate characteristics is 0.2, and the weight of load duration characteristics is 0.1.

[0024] In practical implementation, the distance between the load characteristic vector corresponding to the predicted future power grid load curve and the historical load characteristic vector corresponding to each historical window is calculated. The distance can be calculated using the Euclidean distance formula. The formula is as follows: Where: symbol Represents distance value, symbol The weights representing the peak load characteristics, with the symbol... The symbol represents the characteristic value of the future load curve's peak load. Represents the characteristic value of the peak load within a historical window, with the symbol... The weight representing the load trough characteristic, symbol The characteristic value of the load trough of the future load curve is represented by the symbol. The characteristic value of the load trough in the historical window is represented by the symbol. The weights representing the load change rate characteristics, with symbols... The characteristic value of the load change rate of the future load curve is represented by the symbol. The characteristic value of the load change rate in the historical window is represented by the symbol. The weights representing the load duration characteristics, with symbols... The characteristic value representing the load duration of the future load curve, with the symbol... This represents the load duration characteristic value of the historical window.

[0025] It is understandable that by setting a distance threshold, all historical time windows with distance values ​​less than the threshold are identified as historical load periods similar to the future power grid load characteristics. The distance threshold can be set as a percentile based on the statistical distribution of historical distance values, such as the upper limit of the smallest 10% of all calculated historical distance values. Alternatively, the distance threshold can be set as an absolute value, such as 0.5. When the distance between the load characteristic vector corresponding to the predicted future power grid load curve and the historical load characteristic vector of a certain historical window is calculated to be 0.3, since 0.3 is less than the distance threshold of 0.5, the 24-hour period corresponding to this historical window is identified as a similar historical load period.

[0026] In some embodiments, the traversal process continues until all historical data is covered. For a database containing 365 days of historical data, sliding in one-hour increments will generate a large number of historical windows and calculate the corresponding distance values. After calculation, a sequence of distance values ​​is obtained. By applying a distance threshold filter, a series of historical date or time period identifiers that meet the conditions are finally output, such as "January 15th (all day)" and "March 22nd (all day)". The 24-hour periods represented by these date identifiers are the historical load periods that are similar to the future grid load characteristics. Optionally, when extracting historical load characteristics, the length of the time window is consistent with the coverage duration of the future predicted grid load curve. If the future prediction covers a working day, the historical time window is also set to 24 hours; if the future prediction only covers a peak period, the historical time window is shortened accordingly.

[0027] In one embodiment of the present invention, the process of extracting historical air conditioning operation record units corresponding to each historical load period from historical air conditioning operation data includes the following steps: Establishing a time index containing a start timestamp and an end timestamp for each historical load period. Locating the timestamp corresponding to the operation record unit in the historical air conditioning operation data. Determining whether the timestamp corresponding to the operation record unit falls completely within the time interval defined by the start timestamp and the end timestamp. If it falls completely, the operation record unit is determined to be a valid operation record unit corresponding to the historical load period. Summarizing all valid operation record units corresponding to all historical load periods to form a set of operation records to be analyzed.

[0028] In practical implementation, a substation air conditioning group control energy-saving method based on load forecasting involves extracting historical air conditioning operation record units corresponding to each historical load period from historical air conditioning operation data, and establishing a time index for each historical load period, which includes a start timestamp and an end timestamp. For example, if a selected similar historical load period is "January 15th 00:00:00 to January 15th 23:59:59", then the start timestamp is "January 15th 00:00:00" and the end timestamp is "January 15th 23:59:59". In the historical air conditioning operation data, the timestamp corresponding to the operation record unit is located. The historical air conditioning operation data is stored in a structured form, and each operation record unit contains the status information of the air conditioning equipment and a timestamp accurate to the second, such as "January 15th 08:30:15".

[0029] Determining whether the timestamp corresponding to a runtime record unit falls entirely within the time interval defined by the start and end timestamps requires precise time comparison. A runtime record unit's timestamp is considered to fall entirely within the time interval only if it simultaneously satisfies the condition of being greater than or equal to the start timestamp and less than or equal to the end timestamp. For example, if a runtime record unit's timestamp is "January 15th 08:30:15", its start timestamp is "January 15th 00:00:00", and its end timestamp is "January 15th 23:59:59", then because "January 15th 08:30:15" is greater than or equal to "January 15th 00:00:00" and less than or equal to "January 15th 23:59:59", this runtime record unit is determined to fall entirely within the time interval. In some embodiments, the determination process can be implemented using a logical expression, which is: Where: symbol This indicates the result of the judgment. A true value means the condition is completely met, and a false value means the condition is not completely met. The symbol is... The symbol represents the time stamp corresponding to the running record unit. The timestamp representing the start of a historical load period, symbol The symbol represents the end timestamp of a historical load period. This represents the logical AND operation.

[0030] If all records fall within the specified range, the operating record unit is determined to be a valid operating record unit corresponding to the historical load period. Optionally, for the historical load period of "January 15th", the system will iterate through all operating record units in the air conditioning historical operation database with timestamps between "January 15th 00:00:00" and "January 15th 23:59:59" and mark them all as valid operating record units corresponding to that period. In practice, air conditioning historical operation data may be recorded at a frequency of one record per minute or one record every five minutes. Therefore, for a 24-hour historical load period, hundreds to thousands of valid operating record units may be extracted. The valid operating record units corresponding to all historical load periods are summarized to form the set of operating records to be analyzed. Assuming the system selects three similar historical load periods, namely "January 15th", "March 10th", and "May 20th", then the set of operating records to be analyzed will contain all valid operating record units extracted from these three days.

[0031] In some embodiments, the precision of the time index can be adjusted to accommodate different data granularities. If the timestamps of the historical air conditioning operation data are only accurate to the hour, then the start and end timestamps can also be set to the hour. It is understood that the timestamps corresponding to the operation record units must fall completely within the time interval; partial overlap or exceeding the boundaries are not accepted. For example, an operation record unit with a timestamp of "January 14th 23:59:30" will not be considered a valid operation record unit for the historical load period of "January 15th," even if it only overlaps with the time period of "January 15th" by a few seconds. Optionally, the aggregated set of operation records to be analyzed is a data structure containing multi-dimensional fields such as time, equipment identifier, and operating status, for use in subsequent analysis steps.

[0032] In one embodiment of the present invention, identifying a set of air conditioning control modes that satisfy the temperature control constraints of a substation includes the following steps. Each valid operation record unit in the set of operation records to be analyzed corresponds to a set of state sequences of air conditioning equipment. (See also...) Figure 3The temperature change sequence of the substation's internal environment is extracted from the effective operation record units. This temperature change sequence is compared with a preset substation temperature control constraint range, and effective operation record units whose temperature change sequence remains within the substation's temperature control constraint range throughout are selected and defined as compliant operation record units. Pattern clustering is performed on the state sequences of the air conditioning equipment corresponding to the compliant operation record units, grouping compliant operation record units with state sequence similarity higher than a preset threshold into the same category. The most representative state sequence is extracted from each category to form an air conditioning control mode. All extracted air conditioning control modes constitute an air conditioning control mode set. The control mode with the highest adaptability to the future power grid load curve is selected from the air conditioning control mode set as the basic control framework. This process includes the following steps: For each air conditioning control mode in the air conditioning control mode set, the air conditioning power consumption curve it will generate during the future time period corresponding to the predicted power grid load curve is predicted. The predicted power grid load curve is superimposed with each predicted air conditioning power consumption curve to form multiple predicted total load curves. Based on the power grid smooth operation index, the smoothness score of each predicted total load curve is calculated. From the set of air conditioning control modes, select the air conditioning control mode that gives the highest smoothness score to the corresponding predicted total load curve, and use its structure as the basic control framework.

[0033] In practical implementation, a substation air conditioning group control energy-saving method based on load forecasting involves identifying a set of air conditioning control modes that meet the substation's temperature control constraints. Each valid operation record unit in the set of operation records to be analyzed corresponds to a set of state sequences of air conditioning equipment. The state sequences may include data on compressor start / stop, fan speed, set temperature, and other data that change over time. The temperature change sequence of the substation's internal environment is parsed from the valid operation record units. This temperature change sequence is typically recorded and associated with temperature sensors deployed within the substation using the same timestamp as the operation record unit.

[0034] The temperature change sequence of the substation's internal environment is compared with a preset substation temperature control constraint range, which can be 22°C to 28°C. Valid operation record units whose temperature change sequences fall entirely within this constraint range are selected and defined as compliant operation record units. For example, a valid operation record unit covers the temperature data for the entire day of January 15th, with temperatures fluctuating between 23.5°C and 26.8°C. Since the temperature remains within the 22°C to 28°C range, this record unit is determined to be a compliant operation record unit. Similarly, another record unit with temperatures fluctuating between 21.5°C and 27°C will not be considered a compliant operation record unit because 21.5°C is below the constraint lower limit of 22°C. Pattern clustering is then performed on the state sequences of the air conditioning equipment corresponding to the compliant operation record units. Compliant operation record units with state sequence similarity higher than a preset threshold are grouped into the same category. The preset threshold can be set empirically, for example, a similarity coefficient greater than 0.85.

[0035] In some embodiments, the similarity of state sequences can be calculated using a method based on dynamic time warping (DTW) distance, and the calculation formula is as follows: Where: symbol Represents two state sequences and Similarity coefficient between them, sign Represents a sequence with sequence The dynamic time-warped distance between them. The most representative state sequence is extracted from each class to form an air conditioning control mode. Extracting the most representative state sequence can be done by calculating the average trajectory of all sequences within the class. All extracted air conditioning control modes constitute an air conditioning control mode set. For example, clustering may yield three different operating modes, corresponding to "Control Mode A", "Control Mode B", and "Control Mode C"—three air conditioning control modes.

[0036] In practical implementation, the control mode with the highest adaptability to the future power grid load curve is selected from the set of air conditioning control modes as the basic control framework. For each air conditioning control mode in the set, the power consumption curve of the air conditioning system during the future period corresponding to the predicted power grid load curve is predicted. The prediction can be based on the equipment power model and historical operating data, such as the average power consumption curve of control mode A during similar historical days, which is corrected by the temperature forecast for the future period to serve as its predicted air conditioning power consumption curve for the future period. The predicted power grid load curve is then superimposed with each predicted air conditioning power consumption curve to form multiple predicted total load curves. The superposition is an algebraic addition corresponding to each point in time.

[0037] Based on the power grid smooth operation index, a smoothness score is calculated for each predicted total load curve. The smoothness score can be measured by calculating the sum of squares of the first-order differences (load differences between adjacent time points) of the total load curve; the smaller the sum of squares, the smoother the curve. See Table 1 for a simplified example of smoothness score calculation comparison.

[0038] Table 1: Smoothness Scoring Table of Predicted Total Load Curve for Different Control Modes From the set of air conditioning control modes, the air conditioning control mode that gives the highest smoothness score to the corresponding predicted total load curve is selected, and its structure is used as the basic control framework. Optionally, according to the table above, control mode B has the smallest smoothness score, meaning its predicted total load curve is the smoothest; therefore, the structure of control mode B is selected as the basic control framework. In some embodiments, the basic control framework includes a set of preset operating state instructions for the air conditioning group at each point in time within a future period. It is understood that the calculation method for the smoothness score is not limited to the sum of squared load change rates; the standard deviation of load fluctuations or other mathematical indicators can also be used. Optionally, the prediction of the air conditioning power consumption curve may need to consider the performance degradation factor of the air conditioning equipment, and a correction coefficient based on the equipment's operating years may be introduced into the prediction model.

[0039] In one embodiment of the present invention, the process of identifying adjustable operating periods in the basic control framework includes the following steps: In the predicted air conditioning power consumption curve corresponding to the basic control framework, high-power periods of air conditioning that overlap with the peak load periods in the power grid load curve are identified. Within these high-power periods, based on the redundancy adjustment capability of the air conditioning equipment, sub-periods with the potential to reduce power without affecting temperature control constraints are marked as adjustable operating periods. In the basic control framework, the control command segments corresponding to the adjustable operating periods are highlighted or specially marked. The user's control commands for the adjustable operating periods are coupled and analyzed with the future predicted power grid load curve, including the following steps: The control commands include adjustments to the operating parameters of the air conditioning equipment within a specific adjustable operating period. The control commands are applied to the basic control framework, and the adjusted air conditioning power consumption change and the corresponding temperature change of the substation's internal environment are simulated and calculated. It is determined whether the adjusted temperature change of the substation's internal environment exceeds the substation's temperature control constraint range; if it does, the control commands are corrected. In the predicted power grid load curve, the future power grid load point corresponding to a specific controllable operating period is located. The change in air conditioning power consumption is time-series aligned with the future power grid load point. The impact of the change in air conditioning power consumption on the power grid load fluctuation during the period in which the future power grid load point is located is analyzed, and coupled analysis results including quantitative indicators of the impact are generated.

[0040] In practical implementation, a substation air conditioning group control energy-saving method based on load forecasting involves identifying adjustable operating periods in the basic control framework. The predicted air conditioning power consumption curve corresponding to the basic control framework describes the predicted power value of the air conditioning group at each moment in the future period. The high power consumption period of the air conditioning group that overlaps with the peak load period in the power grid load curve is identified in the predicted air conditioning power consumption curve corresponding to the basic control framework. The peak load period in the power grid load curve is one or more consecutive time periods with the highest power in the future predicted power grid load curve. For example, if the forecast shows that the peak load period is from 14:00 to 16:00 on April 3, and the basic control framework predicts that the power consumption of the air conditioning group is also at a high level during this period, the overlapping period is identified as the high power consumption period of the air conditioning group.

[0041] During periods of high air conditioning power consumption, based on the redundancy adjustment capability of the air conditioning equipment, sub-periods with the potential to reduce power without affecting temperature control constraints are marked as adjustable operating periods. The redundancy adjustment capability of the air conditioning equipment refers to the adjustable range of its operating parameters (such as set temperature and fan speed) while meeting current temperature control requirements. This allows for power reduction within a certain range without causing the ambient temperature to exceed the preset substation temperature control constraint range. For example, in the sub-period from 14:30 to 15:30, simulation calculations show that raising the air conditioning set temperature from 26 degrees Celsius to 27 degrees Celsius can reduce the total power consumption of the air conditioning unit while ensuring the substation indoor temperature remains within the 22-28 degrees Celsius constraint range. Therefore, the sub-period from 14:30 to 15:30 is marked as an adjustable operating period. In the basic control framework, the control instruction segments corresponding to the adjustable operating periods are highlighted or specially marked. These control instruction segments contain the original control parameters of all air conditioning equipment within that period. Special marking can be achieved by adding an "adjustable" label to these instructions in the strategy file or by highlighting them with different colors on the graphical interface.

[0042] In some embodiments, user-issued control commands for adjustable operating periods are coupled with predicted grid load curves. The control commands include adjustments to the operating parameters of air conditioning equipment within a specific adjustable operating period. For example, a user selects the adjustable operating period of 14:30-15:30 on the interface and issues the command "Increase the set temperature of all air conditioners by 1 degree Celsius." The control commands are applied to the basic control framework to simulate and calculate the changes in air conditioning power consumption and the corresponding temperature changes within the substation. This requires simulation calculations using an air conditioning energy consumption model and an indoor thermal balance model. The system determines whether the adjusted temperature changes within the substation exceed the substation's temperature control constraints. If so, the control commands are corrected. For example, if increasing the temperature by 1 degree Celsius would cause the predicted temperature to reach 28.5 degrees Celsius at 15:20, exceeding the 28-degree Celsius upper limit, the system automatically corrects the control command to "Increase by 0.5 degrees Celsius" and recalculates until the temperature prediction meets the constraints.

[0043] In the predicted grid load curve, the future grid load point corresponding to a specific controllable operating period is located. The change in air conditioning power consumption is time-series aligned with the future grid load point. The impact of the change in air conditioning power consumption on the grid load volatility during the time period of the future load point is analyzed, generating a coupled analysis result that includes quantitative indicators of impact. It can be understood that the quantitative indicators of impact can be reflected by calculating the change in the smoothness of the grid load curve before and after adjustment. See Table 2, which shows a simplified example of the coupled analysis result.

[0044] Table 2: Results of Coupling Analysis for Different Control Commands See Figure 4 The process for identifying the basic power consumption curve and adjustable operating periods of air conditioning revolves around load peak adaptation and adjustable space identification. Specifically, the basic predicted power consumption curve (solid line) represents the predicted power value of the substation air conditioning group at each moment in the future under the basic control framework. Its fluctuation trend is highly correlated with the peak-valley characteristics of the future power grid load curve. The high power consumption threshold (dashed line) is the critical standard for determining high power consumption periods, used to filter out high power consumption intervals of air conditioning that overlap with the peak power grid load periods. The identification of adjustable operating periods (filled areas) follows this logic: First, in the basic predicted power consumption curve of air conditioning, locate the continuous interval that overlaps with the peak power grid load periods and has power consumption higher than the high power consumption threshold; second, within this interval, based on the redundancy adjustment capability of the air conditioning equipment, filter out sub-periods with the potential to reduce power while still meeting the substation temperature control constraints after adjustment. Finally, complete the identification with highlighted areas, providing a clear operating range for subsequent user control command issuance and coupling analysis. In the parameter and scenario configuration, the observation window is set from 10:00 to 18:00 on April 3, 2026. The high power consumption threshold of the air conditioner is set to 107kW. The adjustable operating period corresponds to the peak period of the grid load overlap from 14:00 to 15:30. During this period, the peak power consumption of the air conditioner reaches 130kW, which has significant room for power reduction. Moreover, the simulation of the thermal balance model has verified that adjusting the air conditioner operating parameters during this period can achieve load shaving while meeting the temperature control constraint of 22-28℃, which fully meets the judgment criteria for the adjustable operating period.

[0045] In one embodiment of the present invention, the conflict verification model is constructed based on a rule-based reasoning engine. The conflict verification model has a pre-set set of verification rules, including temperature control rules, load fluctuation constraint rules, and equipment start-stop safety interval rules. The conflict verification model compares each control command in the preliminary air conditioning group control execution strategy with all verification rules one by one. When a control command triggers any verification rule, the time period corresponding to that control command is marked as a conflict time period, and the specific verification rule information triggered is recorded. The conflict verification model outputs a verification report containing all conflict time periods and their triggered specific verification rule information. After generating the preliminary air conditioning group control execution strategy, a strategy optimization step is also included. Based on the impact quantification index in the coupling analysis results, the adjustable operating segments in the non-conflict time periods of the preliminary air conditioning group control execution strategy are prioritized. Based on the verification report, control commands in conflict time periods are corrected according to the verification rules they trigger. Combining the priority ranking results and command correction results, an optimization suggestion list is generated, and the optimization suggestion list is merged with the preliminary air conditioning group control execution strategy to form the final air conditioning group control energy-saving operation strategy. The execution of the final air conditioning group control energy-saving operation strategy includes the following steps. The final energy-saving operation strategy for group air conditioning control is decomposed into a discrete set of time-sequential control instructions. A communication link is established between these instructions and the control terminal of the substation's air conditioning system, and the instructions are issued sequentially. During execution, real-time data on the substation's actual temperature and the power grid's actual load are collected. The actual temperature data is compared with the preset substation temperature control constraint range, and the actual load data is compared with the predicted future power grid load curve. When the actual temperature or load data deviates beyond a preset deviation threshold, a strategy regeneration process is triggered, generating a new energy-saving operation strategy for group air conditioning control starting from the current moment.

[0046] In its implementation, the conflict verification model is built upon a rule-based reasoning engine. The model contains a pre-defined set of verification rules, including temperature control rules, load fluctuation constraint rules, and equipment start-stop safety interval rules. The temperature control rules require that after any control command is executed, the predicted internal temperature of the substation must always remain within the preset substation temperature control constraint range, such as 22 to 28 degrees Celsius. The load fluctuation constraint rules require that adjustments to the total power of the air conditioning group should not cause the rate of change of the total load curve after superimposing the grid base load to exceed a set threshold within any short-term window. The equipment start-stop safety interval rules require that there must be at least a ten-minute interval between two start-up operations of the same air conditioning compressor. The conflict verification model compares each control command in the initial air conditioning group control execution strategy with all verification rules one by one. The initial air conditioning group control execution strategy is a list containing timestamps, equipment identifiers, and action commands.

[0047] When a control command triggers any verification rule, the time period corresponding to the control command is marked as a conflict period, and the specific verification rule information is recorded. For example, a control command plans to start an air conditioner compressor at 14:05, but the compressor was last shut down at 13:58, which violates the "ten-minute" safety interval rule. Therefore, 14:05 is marked as a conflict period, and the rule "Equipment start / stop safety interval rule triggered" is recorded. The conflict verification model outputs a verification report containing all conflict periods and their specific triggering verification rule information. The verification report can be a table listing the conflict time, the equipment involved, the type of rule violated, and the original conflict command content.

[0048] In practical implementation, after generating the preliminary air conditioning group control execution strategy, a strategy optimization step is also included. Based on the impact quantification index in the coupling analysis results, the controllable operating segments during non-conflict periods in the preliminary air conditioning group control execution strategy are prioritized. The impact quantification index in the coupling analysis results is, for example, the load fluctuation impact index. The smaller the index value, the more significant the positive effect of the control behavior on smoothing the grid load. A priority score can be calculated for each controllable operating segment. The calculation formula is as follows: Where: symbol Represents the priority score of the adjustable runtime segment, symbol This represents the load fluctuation impact index of the control command given in the coupling analysis results for that period, with the symbol... The symbol represents the expected energy savings resulting from this control order. A very small positive number is used to prevent the denominator from being zero. The higher the score of the controllable operating period, the higher the likelihood that its control instructions will be prioritized or strengthened in the optimization recommendations. In some embodiments, control instructions in conflicting periods are corrected according to the verification rules triggered by the verification report. For example, for instructions that violate the equipment start-stop safety interval rules, the system automatically delays their execution time until the minimum interval requirement is met; for instructions that violate temperature control rules, the system automatically adjusts the temperature setpoint or fan speed to within a safe boundary. Combining the priority ranking results and instruction correction results, an optimization recommendation list is generated and merged with the preliminary air conditioning group control execution strategy to form the final air conditioning group control energy-saving operation strategy. The optimization recommendation list may include "increasing the priority of control instructions during the 14:30-15:30 period" or "changing the compressor start command at 14:05 to be executed at 14:08," etc. The merging process involves replacing conflicting instructions in the original strategy with corrected instructions and sorting and marking non-conflicting controllable instructions according to priority.

[0049] The execution of the final air conditioning group control energy-saving operation strategy can be understood to include the following steps: The final air conditioning group control energy-saving operation strategy is decomposed into a discrete set of time-sequential control instructions. Each instruction specifies the control action for a specific air conditioning unit or group of air conditioning units at a specific moment. A communication link is established between the time-sequential control instruction set and the control terminal of the substation air conditioning system, and the instructions are sent out in chronological order. This communication link can be established via industrial Ethernet or a dedicated wireless network. During execution, real-time data collection is performed on the actual temperature data of the substation and the actual load data of the power grid. The actual temperature data is compared with the preset substation temperature control constraint range, and the actual load data is compared with the predicted future power grid load curve. When the actual temperature data or actual load data is detected to deviate from the preset deviation threshold, the strategy regeneration process is triggered. A new energy-saving operation strategy for the air conditioning group control is generated starting from the current moment. For example, if at 15:00 the actual room temperature is found to have reached 27.8 degrees Celsius and is still rising rapidly, about to reach the upper limit of 28 degrees Celsius, and the original strategy does not react to this, the deviation threshold is exceeded, the system immediately interrupts the execution of the current strategy, and re-executes the complete process from load screening to strategy generation based on the latest data and load forecast, generating a new control strategy starting from the current moment.

[0050] See Figure 5In the substation air conditioning group control energy-saving method based on load forecasting, the comparison between historical and predicted grid load curves is the core data foundation for screening similar load periods and matching subsequent control modes. Specifically, the dashed line represents the historical similar load curve, which is the load curve for a historical period that highly matches the characteristics of future loads after extracting load peak, valley, rate of change, and duration features from historical grid load data, constructing a feature vector, and calculating the vector distance. The solid line represents the future predicted load curve, which is the grid load forecast result for the next 24 hours generated based on grid operation data and meteorological and other influencing factors. The filled area represents the load growth area, intuitively quantifying the incremental space of the future predicted load relative to the historical similar load, providing an intuitive basis for assessing the controllable potential of air conditioning group control. The core technical logic of curve comparison is as follows: First, multi-dimensional load features are extracted from the future predicted load curve. Historical load data is traversed within a fixed time window, and corresponding features are extracted. After constructing a weighted feature vector, the distance between vectors is calculated. Historically similar time periods with distances less than a threshold are selected, and their load curves are represented by the blue dashed lines in the figure. The future predicted load curve is the core input for generating subsequent air conditioning group control strategies. Its load peak, valley, and fluctuation characteristics directly determine the adaptability selection of the air conditioning control mode. The area and distribution of the load growth region can be directly used to assess the growth rate and time distribution of the power grid load, providing data support for identifying controllable operating periods and carrying out load peak shaving and valley filling within the basic control framework. This serves as a prerequisite for subsequent air conditioning power consumption coupling analysis, conflict verification, and strategy optimization. During parameter configuration, the distance threshold for similar load selection is set to 8% of the magnitude of the future load feature vector, the time window step is 15 minutes, and the weights of load peak, valley, rate of change, and duration in the feature vector are configured as 0.35, 0.25, 0.25, and 0.15, respectively, to ensure the accuracy and engineering practicality of similar time period selection.

[0051] The above are merely preferred embodiments of the present invention and are not intended to limit the present invention in any other way. Any person skilled in the art may make changes or modifications to the above-disclosed technical content to create equivalent embodiments that can be applied to other fields. However, any simple modifications, equivalent changes, and modifications made to the above embodiments based on the technical essence of the present invention without departing from the scope of the present invention shall still fall within the protection scope of the present invention.

Claims

1. A substation air conditioning group control energy-saving method based on load forecasting, characterized in that, include: Collect and store historical power grid load data and substation air conditioning historical operation data, wherein the air conditioning historical operation data includes multiple operation record units; Based on the future power grid load curve, several historical load periods with similar characteristics to the future power grid load are selected from the historical power grid load data; From the historical air conditioning operation data, extract the historical air conditioning operation record units corresponding to each historical load period to form a set of operation records to be analyzed; The set of operation records to be analyzed is analyzed to identify a set of air conditioning control modes that meet the temperature control constraints of the substation. The set of air conditioning control modes includes multiple control modes. The control mode with the highest adaptability to the future power grid load curve is selected from the set of air conditioning control modes as the basic control framework. The adjustable runtime segments are identified within the basic control framework; The user's control commands for the controllable operating period are coupled with the predicted power grid load curve to form a coupling analysis result. The coupling analysis result is then integrated with the basic control framework to generate a preliminary air conditioning group control execution strategy. The preliminary air conditioning group control execution strategy is input into a pre-built conflict verification model to verify whether there are any instructions that conflict with temperature control constraints or power grid load requirements. The conflict verification model outputs a verification report containing the conflict period.

2. The energy-saving method for substation air conditioning group control based on load forecasting according to claim 1, characterized in that, Selecting multiple historical load periods from historical power grid load data that are similar to future power grid load characteristics involves the following steps: The load peak characteristics, load valley characteristics, load change rate characteristics, and load duration characteristics are extracted from the predicted future power grid load curve. The historical power grid load data is traversed, and historical load features are extracted using a fixed time window as the step size. A load feature vector is constructed, which is composed of the load peak feature, load valley feature, load change rate feature and load duration feature according to a preset weight; Calculate the distance between the load feature vector corresponding to the predicted future power grid load curve and the historical load feature vector corresponding to each historical window; A distance threshold is set, and all historical time windows with distance values ​​less than the distance threshold are identified as historical load periods similar to the future power grid load characteristics.

3. The energy-saving method for substation air conditioning group control based on load forecasting according to claim 2, characterized in that, Extracting the historical air conditioning operation record unit corresponding to each historical load period from the historical air conditioning operation data includes the following steps: A time index is created for each historical load period, and the time index includes a start timestamp and an end timestamp. In the historical operating data of the air conditioner, locate the time marker corresponding to the operating record unit; Determine whether the timestamp corresponding to the running record unit falls completely within the time interval defined by the start timestamp and the end timestamp; If it falls completely into the range, the operation record unit is determined to be a valid operation record unit corresponding to the historical load period. All valid operation record units corresponding to all historical load periods are aggregated to form the set of operation records to be analyzed.

4. The energy-saving method for substation air conditioning group control based on load forecasting according to claim 3, characterized in that, Identifying the set of air conditioning control modes that satisfy the substation temperature control constraints includes the following steps: Each valid operation record unit in the set of operation records to be analyzed corresponds to a set of state sequences of air conditioning equipment. The temperature change sequence of the substation's internal environment was extracted from the effective operation record unit; The temperature change sequence of the substation's internal environment is compared with the preset substation temperature control constraint range. Valid operation record units whose temperature change sequence is entirely within the substation temperature control constraint range are selected and defined as compliant operation record units. Pattern clustering is performed on the state sequences of the air conditioning equipment corresponding to the compliant operation record units, and compliant operation record units with state sequence similarity higher than a preset threshold are grouped into the same category; Extract the most representative state sequence from each category to form an air conditioning control mode; All extracted air conditioning control modes constitute the air conditioning control mode set.

5. A substation air conditioning group control energy-saving method based on load forecasting according to claim 4, characterized in that, Selecting the control mode with the highest adaptability to the future power grid load curve from the set of air conditioning control modes as the basic control framework includes the following steps: For each air conditioning control mode in the set of air conditioning control modes, predict the air conditioning power consumption curve that it will generate when it is running in the future time period corresponding to the predicted grid load curve. The predicted future power grid load curve is superimposed with each predicted air conditioning power consumption curve to form multiple predicted total load curves. Based on the power grid smooth operation index, the smoothness score of each predicted total load curve is calculated. From the set of air conditioning control modes, select the air conditioning control mode that gives the highest smoothness score to the corresponding predicted total load curve, and use its structure as the basic control framework.

6. A substation air conditioning group control energy-saving method based on load forecasting according to claim 5, characterized in that, Identifying adjustable runtime segments within the basic control framework includes the following steps: In the predicted air conditioning power consumption curve corresponding to the basic control framework, the high power consumption period of the air conditioner that overlaps with the peak load period in the power grid load curve is identified. During the high power consumption period of the air conditioner, based on the redundancy adjustment capability of the air conditioning equipment, the sub-periods with the potential to reduce power without affecting the temperature control constraints are marked as the adjustable operating periods. In the basic control framework, the control instruction segments corresponding to the adjustable runtime segments are highlighted or specially marked.

7. A substation air conditioning group control energy-saving method based on load forecasting according to claim 6, characterized in that, The coupling analysis of user control commands for the controllable operating period with the predicted future grid load curve includes the following steps: The control command includes the adjustment amount of the air conditioning equipment operating parameters within a specific controllable operating period; The control command is applied to the basic control framework to simulate and calculate the change in air conditioning power consumption and the corresponding temperature change in the substation's internal environment after the adjustment. Determine whether the temperature change of the substation's internal environment after adjustment exceeds the substation's temperature control constraint range; if it does, correct the control command. In the predicted future power grid load curve, the future power grid load point corresponding to the specific controllable operating period is located. The change in air conditioning power consumption is time-series aligned with the future power grid load point. The impact of the change in air conditioning power consumption on the power grid load fluctuation during the time period of the future power grid load point is analyzed, and a coupled analysis result containing quantitative indicators of the impact is generated.

8. A substation air conditioning group control energy-saving method based on load forecasting according to claim 7, characterized in that, The conflict verification model is built based on a rule-based reasoning engine; The conflict verification model has a set of pre-set verification rules, which include temperature control rules, load fluctuation constraint rules, and equipment start-up and shutdown safety interval rules. The conflict verification model compares each control command in the preliminary air conditioning group control execution strategy with all verification rules one by one. When a control command triggers any verification rule, the time period corresponding to the control command is marked as a conflict time period, and the specific verification rule information triggered is recorded. The conflict verification model outputs a verification report containing information on all conflict periods and the specific verification rules that triggered them.

9. A substation air conditioning group control energy-saving method based on load forecasting according to claim 8, characterized in that, After generating the initial air conditioning group control execution strategy, the strategy optimization step is also included: Based on the impact quantification index in the coupling analysis results, the adjustable running segments during non-conflict periods in the preliminary air conditioning group control execution strategy are prioritized. Based on the verification report, the control commands that are in conflict periods are corrected according to the verification rules they trigger. By combining the priority ranking results and instruction correction results, an optimization suggestion list is generated, and the optimization suggestion list is merged with the preliminary air conditioning group control execution strategy to form the final air conditioning group control energy-saving operation strategy.

10. A substation air conditioning group control energy-saving method based on load forecasting according to claim 9, characterized in that, The execution of the final air conditioning group control energy-saving operation strategy includes the following steps: The final air conditioning group control energy-saving operation strategy is decomposed into a discrete set of time-sequenced control instructions; Establish a communication link between the timing control instruction set and the control terminal of the substation air conditioning system, and issue the instructions in chronological order; During the execution process, the actual temperature data of the substation and the actual load data of the power grid are collected in real time. The actual temperature data is compared with the preset temperature control constraint range of the substation, and the actual load data is compared with the predicted power grid load curve. When the actual temperature data or actual load data is detected to deviate from the preset deviation threshold, the strategy regeneration process is triggered to generate a new energy-saving operation strategy for the air conditioning group control, starting from the current moment.

Images