An air conditioner load early warning method, device, equipment and medium

Abstract

The present application relates to the field of air conditioning load, and particularly relates to an air conditioning load early warning method, device, equipment and medium. Real-time collected user total power load data, real-time meteorological data and time labels are input into an initial air conditioning load identification model to obtain single-user air conditioning load data. The initial air conditioning load identification model is constructed by using a ridge regression algorithm according to input historical user total power load data, multi-source meteorological parameter data and time characteristic data, and a label is historical air conditioning load data. According to single-user air conditioning load data, power grid real-time operation state data and real-time environmental temperature, a dynamic early warning threshold is obtained. Single-user air conditioning load data is compared with the dynamic early warning threshold in real time, and early warning information is generated when the air conditioning load exceeds the dynamic early warning threshold. Overload and voltage out-of-limit risks caused by concentrated operation of air conditioning load are reduced.

Description

Technical Field

[0001] This invention relates to the field of air conditioning load, and specifically to an air conditioning load early warning method, device, equipment, and medium. Background Technology

[0002] In the current power system, accurate monitoring and analysis of flexible loads (air conditioning load, charging load, etc.) is key to alleviating peak electricity demand pressure, but existing technologies have significant shortcomings.

[0003] Current technologies primarily rely on metering and data acquisition devices such as electricity meters and data collection terminals, as well as resident records, for monitoring and analysis. However, many air conditioning units lack independent metering and data acquisition devices, and air conditioning loads are affected by multiple factors, including industry, temperature, and time of day. Existing models rely solely on temperature-driven models, resulting in poor stability and low recognition accuracy. These issues prevent flexible loads from achieving comprehensive perception and precise control, making it difficult to support the decision-making needs of core business scenarios such as power supply security and load management. Summary of the Invention

[0004] The purpose of this invention is to provide an air conditioning load early warning method, device, equipment and medium, which solves the problems in the prior art.

[0005] This invention is achieved through the following technical solution:

[0006] In a first aspect, embodiments of the present invention provide an air conditioning load early warning method, comprising:

[0007] The real-time collected total user electricity load data, real-time meteorological data and time tags are input into the initial air conditioning load identification model to obtain single-user air conditioning load data. The initial air conditioning load identification model is constructed using the ridge regression algorithm based on the input of historical total user electricity load data, multi-source meteorological parameter data and time feature data, and the tag is historical air conditioning load data.

[0008] Based on the single-user air conditioning load data, real-time power grid operation status data, and real-time ambient temperature, a dynamic early warning threshold is obtained;

[0009] The system compares the individual user air conditioning load data with the dynamic warning threshold in real time, and generates a warning message when the air conditioning load exceeds the dynamic warning threshold.

[0010] Preferably, the initial air conditioning load identification model is constructed through the following steps:

[0011] A training sample set is constructed based on historical total user electricity load data, multi-source meteorological parameter data, and time characteristic data. The multi-source meteorological parameter data includes temperature data, humidity data, and wind speed data, while the time characteristic data includes time data, weekday type data, and seasonal data.

[0012] Based on the training sample set, historical total electricity load data, multi-source meteorological parameter data, and time characteristic data are used as input variables, and historical air conditioning load data is used as labels. The ridge regression algorithm is used to train the model to obtain an initial set of variable coefficients that reflects the relationship between air conditioning load and each input variable.

[0013] An initial air conditioning load identification model is constructed based on the initial variable coefficient set. The expression of the initial air conditioning load identification model is as follows:

[0014] ,

[0015] in, This is the air conditioning load value. For temperature, For humidity, For wind speed, Let i be the i-th time feature variable. For constant terms, to This is the initial set of variable coefficients.

[0016] Preferably, the method further includes:

[0017] The air conditioning load identification results of the initial air conditioning load identification model within the historical monitoring period are back-checked with the actual operating data to obtain the identification accuracy.

[0018] When the recognition accuracy is lower than the preset standard, the model is retrained using the ridge regression algorithm based on the newly added historical user total electricity load data, multi-source meteorological parameter data and time feature data, and the initial variable coefficient set is adjusted to obtain the optimized variable coefficient set.

[0019] The initial air conditioning load identification model is updated based on the optimized set of variable coefficients to obtain the optimized air conditioning load identification model.

[0020] Preferably, obtaining the dynamic early warning threshold based on the single-user air conditioning load data, real-time power grid operating status data, and real-time ambient temperature includes:

[0021] Based on the single-user air conditioning load data, real-time power grid operation status data, and power grid safety operation standards, an initial warning threshold is set, which includes a line overload threshold and a voltage deviation threshold.

[0022] When the real-time ambient temperature is during the high-load period of air conditioning, the line overload threshold is lowered; when the bus voltage deviation is close to the critical state of the power grid, the range of the voltage deviation threshold is narrowed to obtain the dynamic early warning threshold.

[0023] Preferably, the method further includes:

[0024] Based on the single-user air conditioning load data, a visualization chart is generated, which includes a regional air conditioning load overview map, a load heat distribution map, a load trend curve, and a warning threshold adjustment trajectory map.

[0025] Based on the warning information, a highlighted warning icon and warning details are generated, including the abnormal location, abnormal value, scope of impact, and suggested control measures.

[0026] The visualized charts and highlighted warning icons are simultaneously output to the large visualization display screen of the power grid dispatch system.

[0027] Preferably, the real-time meteorological data is obtained from the meteorological data platform through a multi-protocol adapter interface. The multi-protocol adapter interface supports three protocols: IEC 61850, MQTT, and HTTP. A data standardization processing unit is deployed at the interface end to convert the obtained raw meteorological data into a standardized format that the model can recognize.

[0028] Preferably, the method further includes:

[0029] According to the preset daily synchronization frequency, user profile data, total user electricity load data, meteorological data and power grid operation status data are collected from the marketing system, data acquisition system, meteorological data platform and real-time measurement center through the data access module.

[0030] Outliers were removed from the collected multi-source data according to the 3σ criterion, and missing data were filled in using linear interpolation to obtain a cleaned and verified standardized dataset.

[0031] The standardized dataset is transmitted to the data platform, archived and stored according to user-time-data type, and a data storage index is generated.

[0032] Secondly, embodiments of the present invention provide an air conditioning load early warning device, comprising:

[0033] The prediction module is used to input real-time collected user total electricity load data, real-time meteorological data and time labels into the initial air conditioning load identification model to obtain single-user air conditioning load data. The initial air conditioning load identification model is constructed using the ridge regression algorithm based on the input of historical user total electricity load data, multi-source meteorological parameter data and time feature data, and the label is historical air conditioning load data.

[0034] The early warning threshold module is used to obtain a dynamic early warning threshold based on the single-user air conditioning load data, the real-time power grid operation status data, and the real-time ambient temperature.

[0035] The early warning module is used to compare the single-user air conditioning load data with the dynamic early warning threshold in real time, and generate early warning information when the air conditioning load exceeds the dynamic early warning threshold.

[0036] Thirdly, embodiments of the present invention provide an electronic device, including: at least one processor, at least one memory, and computer program instructions stored in the memory, which, when executed by the processor, implement the method of the first aspect described above.

[0037] Fourthly, embodiments of the present invention provide a storage medium storing computer program instructions, which, when executed by a processor, implement the method of the first aspect described above.

[0038] Compared with the prior art, the present invention has the following advantages and beneficial effects:

[0039] By inputting real-time collected total user electricity load data, real-time meteorological data, and time stamps into an initial air conditioning load identification model, single-user air conditioning load data is obtained. This method achieves the separation and extraction of air conditioning load without adding independent metering devices, reducing monitoring costs and expanding the monitoring scope of air conditioning load. Integrating meteorological parameters and time characteristics as model inputs makes the load identification results more consistent with the actual operation patterns of air conditioning systems. A ridge regression algorithm is used to construct the model, and regularization is applied to handle the correlation between multiple variables, improving the model's stability and applicability under different conditions.

[0040] Dynamic early warning thresholds are derived based on individual user air conditioning load data, real-time power grid operating status data, and real-time ambient temperature. A real-time ambient temperature adjustment mechanism is introduced to adjust the early warning threshold, allowing it to adaptively change with temperature. The threshold is lowered when air conditioning load increases to identify risks in advance, and raised when load decreases to avoid invalid warnings. An initial threshold is set in conjunction with the power grid operating status, ensuring that the early warning limits match the current capacity of the power grid and avoiding missed or false alarms caused by fixed thresholds.

[0041] This method compares individual user air conditioning load data with dynamic early warning thresholds in real time, generating an early warning message when the load exceeds the threshold. Real-time comparison ensures continuous and timely monitoring, and the immediate generation of early warning information provides the location of the anomaly and the extent of load exceeding limits, providing operators with a basis for understanding abnormal conditions. This approach issues warnings before a rapid increase in air conditioning load could affect power grid safety, buying time for preventative control measures and reducing the risk of overload and voltage exceedances caused by concentrated air conditioning load operation. Attached Figure Description

[0042] To more clearly illustrate the technical solutions of the exemplary embodiments of the present invention, the accompanying drawings used in the embodiments will be briefly described below. It should be understood that the following drawings only show some embodiments of the present invention and should not be considered as a limitation of the scope. For those skilled in the art, other related drawings can be obtained based on these drawings without creative effort. In the drawings:

[0043] Figure 1 A flowchart illustrating the air conditioning load early warning method provided by the present invention;

[0044] Figure 2 Visual interface for the air conditioning load early warning method provided by the present invention Figure 1 ;

[0045] Figure 3 Visual interface for the air conditioning load early warning method provided by the present invention Figure 2 ;

[0046] Figure 4 Visual interface for the air conditioning load early warning method provided by the present invention Figure 3 ;

[0047] Figure 5 Visual interface for the air conditioning load early warning method provided by the present invention Figure 4 ;

[0048] Figure 6 Visual interface for the air conditioning load early warning method provided by the present invention Figure 5 ;

[0049] Figure 7 This is a schematic diagram of the structure of the air conditioning load early warning device provided by the present invention;

[0050] Figure 8 This is a schematic diagram of the structure of the electronic device provided by the present invention. Detailed Implementation

[0051] To make the objectives, technical solutions, and advantages of the present invention clearer, the present invention will be further described in detail below with reference to the embodiments and accompanying drawings. The illustrative embodiments and descriptions of the present invention are only used to explain the present invention and are not intended to limit the present invention.

[0052] It should be noted that, in this document, relational terms such as "first" and "second" are used merely to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Furthermore, the terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or apparatus. Without further limitations, an element defined by the phrase "comprising..." does not exclude the presence of additional identical elements in the process, method, article, or apparatus that includes said element.

[0053] It should be noted that all actions involving the acquisition of signals, information, or data in this invention are carried out in compliance with the relevant data protection laws and regulations of the locality and with authorization from the owner of the relevant device.

[0054] Example 1

[0055] Please see Figure 1 This invention provides an air conditioning load early warning method, comprising:

[0056] S1. Input the real-time collected user total electricity load data, real-time meteorological data and time label into the initial air conditioning load identification model to obtain single user air conditioning load data. The initial air conditioning load identification model is constructed using the ridge regression algorithm based on the input of historical user total electricity load data, multi-source meteorological parameter data and time feature data, and the label is historical air conditioning load data.

[0057] Specifically, total user electricity load data refers to the overall electricity usage of users collected through metering devices, including the combined power consumption of air conditioning equipment and other electrical equipment. This can be 96 load data points, with one load value collected every 15 minutes, for a total of 96 time points collected daily, measured in kilowatts. Real-time meteorological data refers to external environmental parameters related to electricity load, including at least physical quantities that directly affect the operating status of air conditioning, such as temperature, humidity, and wind speed. Time stamps are time information identifying the moment of data collection, including at least the specific clock time, day of the week, and season, used to distinguish electricity consumption behavior characteristics at different times.

[0058] The initial air conditioning load identification model is a computational model built based on a supervised learning algorithm, used to separate the load component belonging to air conditioning equipment from the total electricity load of users. The model construction process first requires preparing a training dataset. The input includes total electricity load data collected from historical periods, corresponding multi-source meteorological parameter data, and time feature data. The output, i.e., the model's learning label, is the historical air conditioning load data for the same period. Ridge regression is used to train the model on the above data. By introducing an L2 regularization term in the loss function, ridge regression can maintain the stability of model parameters when dealing with multivariate data with multicollinearity, thereby learning the quantitative mapping relationship between input variables and air conditioning load. After training, the model internally forms a parameter set reflecting the contribution of each input variable to the air conditioning load.

[0059] Real-time collected user total electricity load data, real-time meteorological data, and time stamps are input into a pre-constructed initial air conditioning load identification model. The model calculates based on the mapping relationships learned during the training phase, separating the portion of the real-time total electricity load contributed by air conditioning equipment. The output is the individual user air conditioning load data. This process relies on the model's sensitivity analysis to meteorological parameters and time characteristics. Since the air conditioning operating status is closely related to external ambient temperature, humidity, and time period, the model can reasonably estimate the power consumption of air conditioning equipment at the current moment based on current meteorological conditions and time attributes. The advantage of this method is that it does not rely on independent metering devices for air conditioning equipment; it can estimate air conditioning load using only existing user total electricity load data and publicly available meteorological data. This solves the data gap problem caused by the lack of independent monitoring methods for many air conditioning devices. Furthermore, by integrating multi-dimensional input variables, the reliability and applicability of the load decomposition results are improved.

[0060] In some implementations, the initial air conditioning load identification model is constructed through the following steps:

[0061] A training sample set is constructed based on historical total user electricity load data, multi-source meteorological parameter data, and time characteristic data. The multi-source meteorological parameter data includes temperature data, humidity data, and wind speed data, while the time characteristic data includes time data, weekday type data, and seasonal data.

[0062] Specifically, historical total user electricity load data refers to the overall electricity usage records of users collected by metering devices over a past period, including the combined power consumption of air conditioning equipment and other electrical equipment, measured in kilowatts. Multi-source meteorological parameter data refers to records of external environmental parameters corresponding to historical time periods, including temperature, humidity, and wind speed data. These parameters directly affect the operating status and energy consumption levels of air conditioning equipment. Time feature data refers to the time attribute information identifying the historical data collection time, including time data to distinguish different hourly periods within a day, weekday data to distinguish between weekdays and rest days, and seasonal data to distinguish the differences in electricity consumption patterns in spring, summer, autumn, and winter. Constructing a training sample set involves aligning and integrating the above three types of data over time to form a complete dataset containing input features and output labels. Each sample corresponds to a specific time point. The input includes the total user electricity load data, meteorological parameter data, and time feature data for that time point. The output, i.e., the training label for the model, is the historical air conditioning load data for the same period. This sample set provides the basic data support for subsequent model training.

[0063] Based on the training sample set, historical total electricity load data, multi-source meteorological parameter data, and time characteristic data are used as input variables, and historical air conditioning load data is used as labels. The ridge regression algorithm is used to train the model to obtain an initial set of variable coefficients that reflects the relationship between air conditioning load and each input variable.

[0064] Specifically, input variables refer to the set of independent variables used to predict air conditioning load, including historical total electricity load data, multi-source meteorological parameter data, and time characteristic data. These variables are considered the main factors affecting changes in air conditioning load. Labels refer to the target variables during model training, i.e., historical air conditioning load data, representing the true values ​​that the model needs to learn and predict. Ridge regression is a linear regression method suitable for handling multicollinear data. By introducing an L2 regularization term into the standard linear regression loss function, it penalizes the sum of squares of model parameters, thus maintaining the stability of parameter estimation even when there is correlation between multiple input variables. The model training process involves substituting the input variables and labels from the training sample set into the ridge regression algorithm. The algorithm calculates a set of coefficients that optimally balances prediction error and parameter complexity by minimizing the loss function containing the regularization term. This set of coefficients is the initial variable coefficient set, where each coefficient corresponds to an input variable. The magnitude and sign of the coefficient reflect the degree and direction of the variable's influence on air conditioning load. For example, a positive coefficient for temperature indicates that an increase in temperature leads to an increase in air conditioning load, while a negative coefficient for wind speed indicates that an increase in wind speed may reduce air conditioning load.

[0065] An initial air conditioning load identification model is constructed based on the initial variable coefficient set. The expression of the initial air conditioning load identification model is as follows:

[0066] ,

[0067] in, This is the air conditioning load value. For temperature, For humidity, For wind speed, Let i be the i-th time feature variable. For constant terms, to This is the initial set of variable coefficients.

[0068] Specifically, the initial set of variable coefficients is the set of parameters obtained during the model training phase, which includes constant terms. and the coefficients of each input variable to Air conditioning load value This is the model's output prediction, representing an estimate of the air conditioning equipment's power consumption under given input conditions. Temperature T, humidity H, and wind speed W are the basic meteorological parameters input into the model as feature variables. Time feature variables. This represents the encoded time attribute data, such as time of day, week type, and season, which are then numerically transformed and used as independent input dimensions. The model expression uses a linear weighted summation form, multiplying each input variable by its corresponding coefficient and summing the results, then adding a constant term to obtain the predicted value of the air conditioning load. This expression clearly demonstrates the quantitative relationship between the air conditioning load and various influencing factors, with coefficients... This reflects the marginal impact of temperature changes on air conditioning load, and the coefficient... The coefficient reflects the degree of influence of humidity. This reflects the contribution of wind speed, while the coefficients of each time-specific variable capture the inherent differences in electricity consumption patterns across different time periods. The completed initial air conditioning load identification model can be used for subsequent real-time air conditioning load decomposition. By substituting the newly collected total user electricity load data, real-time meteorological data, and time labels into this expression, the corresponding single-user air conditioning load data can be calculated.

[0069] In some embodiments, the method further includes:

[0070] The air conditioning load identification results of the initial air conditioning load identification model within the historical monitoring period are back-checked with the actual operating data to obtain the identification accuracy.

[0071] Specifically, the historical monitoring period refers to a pre-defined time interval used to evaluate model performance, such as the past week or month. The air conditioning load identification result refers to the predicted air conditioning load for a single user calculated by the initial air conditioning load identification model within this historical monitoring period, after inputting the corresponding total user electricity load data, meteorological data, and time stamps for each sampling time. Actual operating data refers to the true air conditioning load value obtained through other reliable methods within the same historical monitoring period, such as power data directly collected from air conditioning equipment with independent metering devices, or air conditioning load records obtained through on-site measurements. Backtracking verification is the process of comparing and analyzing the model identification results with the actual operating data time-by-time, measuring the model's predictive accuracy by calculating the statistical error index between the two. Identification accuracy is a quantitative indicator of model performance, calculated as the proportion of samples with prediction errors within the allowable range to the total number of samples. For example, if a relative error of less than 10% is considered accurate identification, then the accuracy is the proportion of samples meeting this condition. This step provides a quantitative basis for model performance evaluation, and regular backtracking verification allows for an objective understanding of the model's performance level in practical applications.

[0072] When the recognition accuracy is lower than the preset standard, the model is retrained using the ridge regression algorithm based on the newly added historical total user electricity load data, multi-source meteorological parameter data and time characteristic data, and the initial variable coefficient set is adjusted to obtain the optimized variable coefficient set.

[0073] Specifically, the preset standard refers to a pre-defined threshold for recognition accuracy. For example, setting an accuracy of no less than 85% as the minimum acceptable requirement for the model indicates that the current model can no longer meet the application requirements and needs to be optimized and updated. The newly added historical total user electricity load data refers to user electricity records newly collected after the previous monitoring period, containing the latest information on changes in electricity consumption patterns. The newly added multi-source meteorological parameter data refers to meteorological records such as temperature, humidity, and wind speed from the same period as the newly added electricity data. The newly added time feature data refers to the time attribute information corresponding to the newly added data. Retraining the model involves merging the original training sample set with the newly added data to form an expanded training dataset, and then retraining the model using the Ridge Regression algorithm. The Ridge Regression algorithm re-solves for the parameter values ​​that minimize the loss function on the new dataset. Since the newly added data includes recent changes in the relationship between air conditioning load and influencing factors, the trained coefficient set can reflect these new changes. Adjusting the initial set of variable coefficients means replacing the original coefficient values ​​with newly trained coefficient values ​​to form an optimized set of variable coefficients. This set of coefficients updates the weights of each input variable on the air conditioning load, making the model more consistent with the current situation.

[0074] The initial air conditioning load identification model is updated based on the optimized set of variable coefficients to obtain the optimized air conditioning load identification model.

[0075] Specifically, the optimized variable coefficient set is the set of parameters obtained through retraining, containing constant terms and the latest coefficient values ​​for each input variable. Updating the initial air conditioning load identification model means replacing the original coefficient set stored in the model with the optimized variable coefficient set. The computational expression structure of the model remains unchanged, still in the form of a linear weighted summation. The optimized air conditioning load identification model refers to the current version of the model that can be used for subsequent real-time prediction after the coefficient update. This model inherits the original input-output structure and computational logic, but its internal parameters have been adjusted according to the latest data, enabling it to more accurately reflect the relationship between air conditioning load and influencing factors under current conditions. Through this periodic or conditionally triggered model update mechanism, the air conditioning load identification model can adapt to the impact of seasonal changes, evolution of user electricity consumption behavior, and fluctuations in meteorological conditions, maintaining stable identification performance during long-term operation and avoiding the problem of decreased identification accuracy due to model aging.

[0076] S2. Based on the single-user air conditioning load data, the real-time power grid operation status data, and the real-time ambient temperature, obtain the dynamic early warning threshold;

[0077] Specifically, single-user air conditioning load data is the power consumption value of air conditioning equipment separated from the total electricity load of a user, reflecting the intensity of air conditioning use by a single user at a specific time. Real-time power grid operating status data refers to technical parameters describing the current operating status of the power grid, including at least information characterizing the safe and stable state of the power grid such as line transmission power, bus voltage level, and equipment load rate. Real-time ambient temperature refers to the current external atmospheric temperature value, a key external factor affecting changes in air conditioning load.

[0078] Dynamic early warning thresholds refer to the boundary values ​​used to determine whether air conditioning load is at an abnormal or high-risk level. These threshold values ​​are not fixed but are adaptively adjusted based on real-time operating conditions. The process of obtaining dynamic early warning thresholds first requires setting a set of basic early warning thresholds based on the technical standards for safe operation of the power grid, combined with the distribution characteristics of individual user air conditioning load data and real-time power grid operating status data. For example, an overload warning limit is set for line transmission power, and a deviation allowable range is set for bus voltage. Then, real-time ambient temperature is introduced as an adjustment factor to correct the above basic thresholds. There is a clear physical relationship between ambient temperature and air conditioning load; increased temperature directly leads to increased air conditioning cooling load, while decreased temperature leads to increased heating load. When the real-time ambient temperature is in the high-incidence range of air conditioning load, it means that the possibility of concentrated operation of air conditioning equipment increases during the current period, and the pressure on the power grid increases accordingly. At this time, appropriately lowering the line overload warning threshold can identify potential risks in advance before the load further increases. Similarly, when the power grid operating status data reflects that the bus voltage is approaching the stable operating boundary, narrowing the range of the voltage deviation warning threshold can more sensitively capture the risk of voltage exceeding limits. By combining real-time ambient temperature and power grid operating status into a dynamic threshold adjustment mechanism, the early warning threshold can change with changes in external conditions and system status, avoiding the risk of invalid early warnings during low-load periods or missed warnings during high-load periods when using fixed thresholds, thus improving the adaptability of the early warning system to the actual operating environment.

[0079] In some implementations, S2, based on the single-user air conditioning load data, real-time power grid operating status data, and real-time ambient temperature, obtains a dynamic early warning threshold, including:

[0080] S21. Based on the single-user air conditioning load data, the real-time power grid operation status data, and the power grid safety operation standards, set an initial warning threshold, which includes a line overload threshold and a voltage deviation threshold.

[0081] Specifically, single-user air conditioning load data refers to the power consumption value of air conditioning equipment separated from the total electricity load of a user, reflecting the intensity of air conditioning use by a single user at a specific time. Real-time power grid operation status data refers to technical parameters describing the current operating status of the power grid, including at least information characterizing the safe and stable state of the power grid such as line transmission power, bus voltage level, and equipment load rate. Power grid safe operation standards refer to the technical specifications that the power system must adhere to, which stipulate mandatory indicators such as normal load limits for lines and transformers, permissible voltage deviation ranges, and equipment thermal stability limits. These indicators are the fundamental basis for ensuring the safe and stable operation of the power grid.

[0082] The initial warning threshold refers to a preliminary judgment limit value set based on current operating conditions and safety standards, used to identify potential power grid anomalies caused by air conditioning loads. The process of setting the initial warning threshold first analyzes the proportion and distribution characteristics of individual user air conditioning load data within the overall power grid load. This is combined with the current line load rate and measured bus voltage values ​​from real-time power grid operating status data, and refers to equipment limits and voltage ranges specified in power grid safety operation standards to calculate a warning limit suitable for the current operating conditions. The line overload threshold is one type of initial warning threshold, referring to the upper limit of the load rate set for transmission lines or distribution transformers. When the actual load rate exceeds this threshold, it indicates that the equipment may face overload risk. The voltage deviation threshold is another type of initial warning threshold, referring to the upper and lower limits of the allowable fluctuation range set for bus voltage. When the actual voltage exceeds this range, it indicates abnormal voltage quality or that system stability is affected. By setting initial warning thresholds that include both line overload and voltage deviation, the risks caused by air conditioning loads can be monitored from two dimensions: transmission capacity and power quality, providing a basic reference value for subsequent dynamic adjustments.

[0083] S22. When the real-time ambient temperature is during the high-load period of air conditioning, the line overload threshold is lowered; when the bus voltage deviation is close to the critical state of the power grid, the range of the voltage deviation threshold is narrowed to obtain the dynamic early warning threshold.

[0084] Specifically, high-load periods for air conditioning refer to the periods when air conditioning equipment operates intensively due to excessively high or low ambient temperatures. During the high-temperature periods of summer, the cooling load of air conditioning increases significantly, and during the low-temperature periods of winter, the heating load of air conditioning rises significantly. Both of these periods fall under the category of high-load periods for air conditioning. Real-time ambient temperatures falling within these high-load periods mean that current external temperature conditions will drive a large number of air conditioning devices to operate simultaneously, potentially causing a rapid increase in the overall load level of the power grid and a corresponding increase in the load pressure on local lines and transformers. Lowering the line overload threshold means appropriately reducing its value based on the initial line overload threshold. For example, lowering the threshold from 80% to 75% allows for earlier warnings to be triggered before the line load reaches the thermal stability limit of the equipment. This reduction is based on the prediction that concentrated air conditioning operation may lead to rapid load growth. By lowering the threshold, the sensitivity of the warning is improved, allowing more response time before the actual load exceeds the equipment capacity.

[0085] Bus voltage deviation approaching the grid critical state means that, based on real-time grid operation data, the bus voltage amplitude has approached the permissible deviation boundary specified by the grid safety operation standards. For example, if the standard stipulates that the voltage deviation should not exceed 7% of the rated voltage, and the current measured deviation has reached 6.5%, the system's ability to withstand disturbances decreases, and any additional load change may push the voltage out of the permissible range. Narrowing the voltage deviation threshold range means tightening its upper and lower limits based on the initial voltage deviation threshold. For example, reducing the original threshold range of ±7% to ±6% makes voltage monitoring more sensitive in detecting the risk of exceeding limits. When the voltage approaches the critical state, narrowing the threshold range can provide early warning when the voltage has not yet exceeded the limit but the trend is obvious, giving operators more time to make decisions on voltage adjustment measures.

[0086] The dynamic early warning threshold is a judgment boundary value determined after adjustments based on both real-time ambient temperature and grid operating status. This threshold is not fixed but continuously updated according to changes in ambient temperature and grid conditions. During periods of high air conditioning load, the threshold is lowered to improve overload warning sensitivity; when voltage approaches a critical state, the threshold range is tightened to improve voltage exceedance warning sensitivity; and under normal conditions, it returns to the initial threshold level. Through this dynamic adjustment mechanism, the early warning threshold can adaptively match the risk characteristics of different operating stages. During low-risk periods, it avoids excessive invalid warnings interfering with operational monitoring; during high-risk periods, it strengthens early warning capabilities to promptly identify potential threats, enabling the air conditioning load monitoring and early warning system to better adapt to the actual operating needs of the power grid.

[0087] S3. The single-user air conditioning load data is compared with the dynamic early warning threshold in real time, and an early warning message is generated when the air conditioning load exceeds the dynamic early warning threshold.

[0088] Specifically, real-time comparison refers to the operation of comparing the calculated single-user air conditioning load value at the current moment with the corresponding dynamic warning threshold. This comparison operation is continuously executed at a preset time frequency to ensure continuous monitoring of changes in air conditioning load. An air conditioning load exceeding the dynamic warning threshold means that the single-user air conditioning load data is greater than the upper limit specified by the dynamic warning threshold, or lower than the lower limit specified by the dynamic warning threshold, depending on whether the warning type is for overload risk or low load anomaly.

[0089] Early warning information is suggestive data generated after detecting that the air conditioning load exceeds the dynamic early warning threshold. It is used to convey relevant information about abnormal events to power grid dispatching or operation management personnel. Early warning information includes at least the user identifier of the abnormality, the time of occurrence, the actual load value, the type of trigger threshold, and the extent of load exceeding the limit. After generating the early warning information, it is pushed to the power grid dispatching system or operation monitoring platform for operators to view and handle. This comparison and early warning mechanism enables real-time monitoring of the air conditioning load operation status. When a single user's air conditioning load rapidly increases and exceeds the dynamically adjusted early warning threshold due to abnormal weather, equipment failure, or concentrated use, the system can identify the abnormal event and generate an early warning immediately, providing a basis for operators to take early intervention measures. By combining dynamic thresholds with real-time comparison, the early warning mechanism can adapt to load change patterns under different operating conditions while maintaining a sensitive ability to identify abnormal states, helping to issue warnings before the rapid increase in air conditioning load poses a real impact on power grid security.

[0090] In some embodiments, the method further includes:

[0091] Based on the single-user air conditioning load data, a visualization chart is generated, which includes a regional air conditioning load overview map, a load heat distribution map, a load trend curve, and a warning threshold adjustment trajectory map.

[0092] Specifically, single-user air conditioning load data is the power consumption value of each user at a specific time, obtained after model decomposition. This data can be aggregated according to the user's region to form load information at different spatial scales. Visual charts are a technical means to transform abstract numerical data into a graphical display, making it easier for operators to intuitively understand the load distribution and variation patterns.

[0093] A regional air conditioning load overview map is a macro-level chart that presents the total air conditioning load within each administrative region or power supply zone in numerical or graphical form. It includes at least the ranking of air conditioning load in each district / county, the total load value, and the proportion of the total regional electricity load, enabling operators to quickly grasp the overall distribution pattern of air conditioning load across the entire region. A load heat distribution map is a spatial distribution chart that uses a geographic map as a background and represents the air conditioning load density at each geographic location with different color depths. Areas with higher loads have darker colors, creating a visual effect similar to heat distribution, allowing operators to intuitively identify hotspots of concentrated air conditioning load. A load trend curve is a time-series chart that plots a curve with time on the horizontal axis and load value on the vertical axis. It includes at least a comparison of the total electricity load curve and the air conditioning load curve, as well as the trajectory of air conditioning load changes over different time periods, enabling operators to analyze peak load periods and growth trends. The warning threshold adjustment trajectory chart is a visual representation of the dynamic changes in thresholds. Using time as the horizontal axis and threshold value as the vertical axis, it plots curves showing the changes in line overload thresholds and voltage deviation thresholds as adjusted for real-time ambient temperature and grid conditions. This allows operators to trace the history of threshold adjustments and current settings. By generating these four types of visualization charts, single-user air conditioning load data is presented in a multi-dimensional and multi-format manner, providing intuitive information support for operation monitoring and decision analysis.

[0094] Based on the warning information, a highlighted warning icon and warning details are generated, including the abnormal location, abnormal value, scope of impact, and suggested control measures.

[0095] Specifically, early warning information is suggestive data generated after detecting that the air conditioning load exceeds the dynamic early warning threshold. It is used to convey relevant information about abnormal events to power grid dispatching or operation management personnel. Highlighted early warning indicators are graphic elements that prominently display the early warning status on the visual interface. For example, marking the area or equipment where the warning occurs as flashing red, or adding a warning icon to the corresponding position on the chart, allows operators to quickly locate the abnormal location among a large amount of information.

[0096] The warning details are textual information that provides a detailed description of the warning event, including a comprehensive explanation of the abnormal situation. The abnormal location refers to the specific geographical or equipment location where the warning occurred, including at least the user name, the power supply line, the substation, and administrative division information, enabling operators to accurately pinpoint the location of the abnormality. The abnormal value refers to the actual load measurement that triggered the warning and its comparison with the warning threshold, including at least the current air conditioning load value, the threshold type that triggered the warning, and the extent to which the threshold was exceeded, allowing operators to understand the severity of the abnormality. The impact range refers to the area or equipment that the abnormality may affect, including at least other users on the same line, upstream power sources, and downstream power supply areas, allowing operators to assess the potential chain reactions caused by the abnormality. Recommended control measures are disposal suggestions generated based on the abnormality type and the current grid status, including at least actionable solutions such as peak-shifting power guidance, equipment output adjustment, and reserve capacity deployment, providing a reference for operators to take intervention measures. By generating highlighted warning icons and complete warning details, the warning information is presented in a visually appealing and comprehensive manner, facilitating rapid response and handling of abnormal events by operators.

[0097] The visualized charts and highlighted warning icons are simultaneously output to the large visualization display screen of the power grid dispatch system.

[0098] Specifically, the visualization charts include various graphical displays such as regional air conditioning load overview maps, load heat distribution maps, load trend curves, and early warning threshold adjustment trajectory maps. Highlighted early warning indicators are eye-catching warning elements superimposed on the visualization charts, used to highlight the location and status of the warning. The large-screen visualization display of the power grid dispatching system refers to a large-size display screen or display wall deployed in the dispatching center, used to centrally display the power grid's operating status and monitoring information, allowing dispatchers to monitor and direct operations in real time.

[0099] Synchronous output refers to pushing visual charts and highlighted warning indicators to the large display screen at the same time, maintaining the correspondence between the charts and warning indicators. For example, a red flashing indicator of the warning area is simultaneously displayed on the load heat distribution map, and the location of the warning occurrence is simultaneously marked on the load trend curve. After being output to the large display screen, the visual charts are presented in full-screen or split-screen mode. Dispatchers can simultaneously monitor multiple dimensions of air conditioning load information and directly see the indicators and locations of warning events in the charts. This integrated display method combines air conditioning load monitoring results with warning information, allowing dispatchers to obtain complete abnormal event information without switching interfaces or querying additional data, improving the efficiency of operation monitoring and the timeliness of abnormal response. Through the centralized presentation on the large visual display screen, air conditioning load monitoring results become an integral part of the daily operation monitoring of the power grid, providing intuitive information support for dispatching decisions.

[0100] In some implementations, the real-time meteorological data is obtained from the meteorological data platform through a multi-protocol adapter interface. The multi-protocol adapter interface supports three protocols: IEC 61850, MQTT, and HTTP. A data standardization processing unit is deployed at the interface end to convert the obtained raw meteorological data into a standardized format that the model can recognize.

[0101] In some embodiments, the method further includes:

[0102] According to the preset daily synchronization frequency, user profile data, total user electricity load data, meteorological data and power grid operation status data are collected from the marketing system, data acquisition system, meteorological data platform and real-time measurement center through the data access module.

[0103] Specifically, the preset daily synchronization frequency refers to the pre-defined time interval or specific time for executing data collection tasks each day. For example, it might be set to perform a full data synchronization at 2:00 AM daily, or an incremental data synchronization every hour. This frequency is determined based on business needs and the data update cycle. The data access module is a software functional unit deployed in the server cluster, responsible for establishing connections with external data sources and performing data acquisition operations. This module is configured with the interface parameters and communication protocols corresponding to each data source.

[0104] The marketing system refers to the business system used by the power grid company to manage basic user information. The user profile data stored in it includes at least user ID, user name, electricity address, electricity category, power line number, and equipment capacity. This data is used to identify user identity and electricity usage attributes. The data acquisition system refers to a system composed of metering devices and data acquisition terminals deployed at electricity consumption sites. It is responsible for collecting and uploading total user electricity load data, which is a sequence of power values ​​recorded at set time intervals, in kilowatts. The meteorological data platform refers to an external system that provides meteorological information services. The meteorological data obtained from it includes at least environmental parameters related to air conditioning load, such as temperature, humidity, and wind speed. The real-time measurement center refers to a system platform that collects real-time monitoring data of power grid operation. The power grid operation status data obtained from it includes at least technical parameters reflecting the current operating status of the power grid, such as line transmission power, bus voltage, equipment load rate, and switch status. By executing daily synchronous data acquisition tasks, raw data from various source systems are periodically aggregated to the data access module, providing a basic data source for subsequent data processing and load analysis.

[0105] Outliers were removed from the collected multi-source data according to the 3σ criterion, and missing data were filled in using linear interpolation to obtain a cleaned and verified standardized dataset.

[0106] Specifically, the 3σ criterion is an outlier identification method based on the assumption of a normal distribution. This criterion states that, assuming data follows a normal distribution, the probability of a value falling within the range of the mean plus or minus three standard deviations is 99.73%. Data points outside this range are considered outliers. When applying the 3σ criterion to remove outliers from multi-source collected data, the mean and standard deviation of each data set are first calculated. Then, data points deviating from the mean by more than three standard deviations are identified and removed. These outliers may originate from metering device malfunctions, communication interference, or recording errors; failure to remove them will affect the accuracy of subsequent analysis.

[0107] Linear interpolation is a method for completing missing data, suitable for situations where individual data points are missing in time series data. When missing values ​​exist in the collected data sequence, the estimated value of the missing location is calculated linearly according to the time interval ratio of the adjacent valid data points before and after the missing point. For example, in a user's total electricity load data sequence, if data is missing at a certain moment, the two valid load values ​​before and after that moment are taken, and assuming that the load changes linearly with time, the load value at the midpoint is calculated as the completed data. This method is computationally simple and can maintain the continuity of the data sequence, making it suitable for physical quantities such as load data that change smoothly over time.

[0108] A cleaned and validated standardized dataset refers to the final dataset after outlier removal and missing data completion, and after standardizing the data format and units. Standardization processes include aligning timestamps from various data sources to the same time zone, converting data of different units to a unified unit, and encoding text data into numerical data. This dataset is complete, continuous, and formatted consistently, and can be directly used for subsequent load decomposition and model calculations, avoiding interference from the quality of the original data in the analysis results.

[0109] The standardized dataset is transmitted to the data platform, archived and stored according to user-time-data type, and a data storage index is generated.

[0110] Specifically, the data middle platform is a unified data service platform built by power grid companies, possessing data aggregation, storage, computing, and service capabilities, providing standardized data support for various business applications. After the standardized datasets have been cleaned and verified, they are transmitted to the storage nodes of the data middle platform through the internal network and enter the unified data resource pool.

[0111] Categorized archiving by user-time-data type refers to a storage method that establishes a three-level directory structure within the data platform. The user level uses the user ID as the first-level classification identifier, centralizing all data belonging to the same user. The time level uses the data collection date or time as the second-level classification identifier, organizing data from different times for the same user in chronological order. The data type level uses the data source or content attribute as the third-level classification identifier, storing different types of data from the same user at the same time separately; for example, user profile data, total user electricity load data, meteorological data, and power grid operation status data are stored independently. This three-level classification structure facilitates subsequent data retrieval and access by user, time, and type.

[0112] A data storage index is a catalog system established to accelerate data retrieval. This index records at least the storage location of each data entry, the user to whom the data belongs, the data's timestamp, and its data type identifier. After generating the data storage index, subsequent load analysis or visualization modules do not need to traverse the entire storage space when accessing data; they can simply locate the target data using the index and read it directly. By establishing a storage index, the management and access efficiency of large-scale data is improved, providing data access support for real-time monitoring and historical analysis.

[0113] Furthermore, this embodiment also includes:

[0114] Core hardware support: Relying on an intranet server cluster (including data storage nodes, algorithm computing nodes, and visualization display nodes), computing resources, storage resources, and backup communication links are reserved for each node. A dynamic resource scheduling module is configured so that when the load rate of any node exceeds the threshold, the task is automatically distributed to the backup node to ensure hardware stability. Through a high-speed network, the server cluster establishes point-to-point connections with the "Dynamic and Static Power Grid Map" platform, the data middle platform, and the real-time measurement center to achieve data interoperability.

[0115] The software modules consist of: a core algorithm module for air conditioning load modeling and an adaptive control engine deployed at the algorithm computation node; a data access module deployed at the data storage node; and a visualization monitoring module and an early warning push module deployed at the visualization display node. Each module is initialized and configured, and default operating parameters are set. Through standardized API interfaces, the output of the data access module is connected to the input of the core algorithm module for air conditioning load modeling, and the output of the core algorithm module is connected to the input of the adaptive control engine. The output of the adaptive control engine is connected to the inputs of the visualization monitoring module and the early warning push module, respectively. Simultaneously, the feedback ends of the visualization monitoring module and the early warning push module are connected to the adaptive control engine, forming a complete closed-loop data flow.

[0116] External Interface Design: A multi-protocol adapter interface is designed, supporting three mainstream protocols: IEC 61850, MQTT, and HTTP. These interfaces connect to provincial real-time measurement systems, meteorological data platforms, and power grid dispatching systems, respectively. The interfaces are hot-swappable and adaptable to different system interface specifications. A data standardization processing unit is deployed at the external interface to convert internal product data into a format compatible with the existing digital architecture of the power grid, outputting standardized data reports and warning signals containing air conditioning load data, control suggestions, and early warning information. An interface firewall and intrusion detection system are configured to perform security verification on externally accessed data, allowing only authorized IP addresses to access it. Simultaneously, the output data undergoes de-identification processing to hide sensitive user information, ensuring power grid data security.

[0117] The following specific examples illustrate this embodiment.

[0118] 1. Real-time panoramic monitoring of flexible equipment load

[0119] 1.1 Overview of Flexible Loads

[0120] like Figure 2 As shown, by accessing the original load data and using the ridge regression algorithm, the original load data is combined with the threshold of the variable coefficient set (temperature, humidity, wind speed, time period, etc.) and substituted into the historical data model for backtracking verification to obtain flexible load operation data. The system monitors the number of flexible load users, electricity consumption, number of monitored electricity consumption categories, maximum load and occurrence time, and the proportion of flexible load by type in Chengdu, thus achieving panoramic monitoring and display of flexible loads in the city.

[0121] 1.2 Real-time load monitoring distribution

[0122] like Figure 3 As shown, by accessing the profile data of flexible load users, including user electricity consumption category, district / county, electricity consumption type, and 96 load data points of users, the data is statistically displayed using a panoramic map of Chengdu, including the distribution statistics of total load, flexible load, and flexible load percentage, and different colors are used to distinguish the load situation at each level.

[0123] 1.3 Flexible Load Trend Analysis and Display

[0124] like Figure 4 As shown, by accessing 96 load data points from users, the system displays the load trend by type (total power load / low-voltage flexible load / high-voltage flexible load), showing the peak and off-peak periods of user power consumption, providing data support for demand response and load regulation across the city.

[0125] 1.4 Industry Flexible Load Analysis and Display

[0126] like Figure 5As shown, by accessing flexible user profile data and load data from 96 points, the system monitors the flexible load of various industries on a daily basis, including ranking of flexible load by industry and ranking of flexible load as a percentage of total load, thus providing data support for demand response and load regulation across the city.

[0127] 1.5 Flexible Load Analysis and Display for Districts and Counties

[0128] like Figure 6 As shown, by accessing flexible user profile data and load data from 96 points, the load situation of each district and county in Chengdu is monitored and statistically analyzed on a daily basis, including the ranking of flexible load in districts and counties and the ranking of the proportion of flexible load to total load, so as to provide data support for demand response and load regulation in the city.

[0129] Example 2

[0130] Please see Figure 7 This invention provides an air conditioning load early warning device, comprising:

[0131] Prediction module 201 is used to input real-time collected user total electricity load data, real-time meteorological data and time labels into an initial air conditioning load identification model to obtain single-user air conditioning load data. The initial air conditioning load identification model is constructed using ridge regression algorithm based on the input of historical user total electricity load data, multi-source meteorological parameter data and time feature data, and the label is historical air conditioning load data.

[0132] The early warning threshold module 203 is used to obtain a dynamic early warning threshold based on the single-user air conditioning load data, the real-time power grid operation status data, and the real-time ambient temperature.

[0133] The early warning module 204 is used to compare the single-user air conditioning load data with the dynamic early warning threshold in real time, and generate early warning information when the air conditioning load exceeds the dynamic early warning threshold.

[0134] It should be noted that each module and unit in the air conditioning load early warning device in this embodiment corresponds one-to-one with each step in the air conditioning load early warning method in the aforementioned embodiment. Therefore, the specific implementation of this embodiment can refer to the implementation of the aforementioned air conditioning load early warning method, and will not be repeated here.

[0135] Example 3

[0136] Please see Figure 8 This embodiment provides an electronic device, including at least one processor 301 and a memory 302. Optionally, the device further includes a communication component 303. The processor 301, memory 302, and communication component 303 are connected via a bus 304.

[0137] In a specific implementation, at least one processor 301 executes computer execution instructions stored in memory 302, causing at least one processor 301 to perform the above-described method.

[0138] The specific implementation process of processor 301 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.

[0139] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.

[0140] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.

[0141] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.

[0142] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.

[0143] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.

[0144] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.

[0145] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.

[0146] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.

[0147] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.

[0148] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.

[0149] If a function is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0150] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.

[0151] The specific embodiments described above further illustrate the purpose, technical solution, and beneficial effects of the present invention. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc., made within the spirit and principles of the present invention should be included within the scope of protection of the present invention.

Claims

1. A method for early warning of air conditioning load, characterized in that, include: The real-time collected total user electricity load data, real-time meteorological data and time tags are input into the initial air conditioning load identification model to obtain single-user air conditioning load data. The initial air conditioning load identification model is constructed using the ridge regression algorithm based on the input of historical total user electricity load data, multi-source meteorological parameter data and time feature data, and the tag is historical air conditioning load data. Based on the single-user air conditioning load data, real-time power grid operation status data, and real-time ambient temperature, a dynamic early warning threshold is obtained; The system compares the individual user air conditioning load data with the dynamic warning threshold in real time, and generates a warning message when the air conditioning load exceeds the dynamic warning threshold.

2. The method according to claim 1, characterized in that, The initial air conditioning load identification model is constructed through the following steps: A training sample set is constructed based on historical total user electricity load data, multi-source meteorological parameter data, and time characteristic data. The multi-source meteorological parameter data includes temperature data, humidity data, and wind speed data, while the time characteristic data includes time data, weekday type data, and seasonal data. Based on the training sample set, historical total electricity load data, multi-source meteorological parameter data, and time characteristic data are used as input variables, and historical air conditioning load data is used as labels. The ridge regression algorithm is used to train the model to obtain an initial set of variable coefficients that reflects the relationship between air conditioning load and each input variable. An initial air conditioning load identification model is constructed based on the initial variable coefficient set. The expression of the initial air conditioning load identification model is as follows: ， in, This is the air conditioning load value. For temperature, For humidity, For wind speed, Let i be the i-th time feature variable. For constant terms, to This is the initial set of variable coefficients.

3. The method according to claim 2, characterized in that, The method further includes: The air conditioning load identification results of the initial air conditioning load identification model within the historical monitoring period are back-checked with the actual operating data to obtain the identification accuracy. When the recognition accuracy is lower than the preset standard, the model is retrained using the ridge regression algorithm based on the newly added historical total user electricity load data, multi-source meteorological parameter data and time characteristic data, and the initial variable coefficient set is adjusted to obtain the optimized variable coefficient set. The initial air conditioning load identification model is updated based on the optimized set of variable coefficients to obtain the optimized air conditioning load identification model.

4. The method according to claim 1, characterized in that, The process of obtaining a dynamic early warning threshold based on the single-user air conditioning load data, real-time power grid operation status data, and real-time ambient temperature includes: Based on the single-user air conditioning load data, real-time power grid operation status data, and power grid safety operation standards, an initial warning threshold is set, which includes a line overload threshold and a voltage deviation threshold. When the real-time ambient temperature is during the high-load period of air conditioning, the line overload threshold is lowered; when the bus voltage deviation is close to the critical state of the power grid, the range of the voltage deviation threshold is narrowed to obtain the dynamic early warning threshold.

5. The method according to claim 1, characterized in that, The method further includes: Based on the single-user air conditioning load data, a visualization chart is generated, which includes a regional air conditioning load overview map, a load heat distribution map, a load trend curve, and a warning threshold adjustment trajectory map. Based on the warning information, a highlighted warning icon and warning details are generated, including the abnormal location, abnormal value, scope of impact, and suggested control measures. The visualized charts and highlighted warning icons are simultaneously output to the large visualization display screen of the power grid dispatch system.

6. The method according to claim 1, characterized in that, The real-time meteorological data is obtained from the meteorological data platform through a multi-protocol adapter interface. The multi-protocol adapter interface supports three protocols: IEC 61850, MQTT, and HTTP. A data standardization processing unit is deployed at the interface end to convert the acquired raw meteorological data into a standardized format that the model can recognize.

7. The method according to claim 1, characterized in that, The method further includes: According to the preset daily synchronization frequency, user profile data, total user electricity load data, meteorological data and power grid operation status data are collected from the marketing system, data acquisition system, meteorological data platform and real-time measurement center through the data access module. Outliers were removed from the collected multi-source data according to the 3σ criterion, and missing data were filled in using linear interpolation to obtain a cleaned and verified standardized dataset. The standardized dataset is transmitted to the data platform, archived and stored according to user-time-data type, and a data storage index is generated.

8. An air conditioning load early warning device, characterized in that, include: The prediction module is used to input real-time collected user total electricity load data, real-time meteorological data and time labels into the initial air conditioning load identification model to obtain single-user air conditioning load data. The initial air conditioning load identification model is constructed using the ridge regression algorithm based on the input of historical user total electricity load data, multi-source meteorological parameter data and time feature data, and the label is historical air conditioning load data. The early warning threshold module is used to obtain a dynamic early warning threshold based on the single-user air conditioning load data, the real-time power grid operation status data, and the real-time ambient temperature. The early warning module is used to compare the single-user air conditioning load data with the dynamic early warning threshold in real time, and generate early warning information when the air conditioning load exceeds the dynamic early warning threshold.

9. An electronic device, characterized in that, include: At least one processor, at least one memory, and computer program instructions stored in the memory, which, when executed by the processor, implement the method as described in any one of claims 1-7.

10. A computer-readable storage medium having computer program instructions stored thereon, characterized in that, The method as described in any one of claims 1-7 is implemented when the computer program instructions are executed by the processor.

Images