CN-122000953-A - Intelligent conservation method for compressed air energy storage system power grid based on big data

CN122000953ACN 122000953 ACN122000953 ACN 122000953ACN-122000953-A

Abstract

The invention relates to the technical field of data processing, in particular to an intelligent daemon of a power grid of a compressed air energy storage system based on big data, which comprises the steps of collecting operation data in the power grid of the compressed air energy storage system, cleaning, denoising and extracting characteristics of the operation data to obtain operation characteristics; when the prediction result is at risk, the compressed air energy storage system is switched to a power grid daemon hot standby mode from a shutdown state or a normal standby state, whether the length of a moving average window needs to be increased or not is determined, the transmission delay time of operation data is acquired to determine whether the suitability of the prediction model update meets the requirement or not, a dynamic threshold of the minimum training data amount is determined whether the minimum training data amount needs to be reduced or not, and if the dynamic threshold of the minimum training data amount does not need to be reduced, the offset of the update triggering time is determined based on the ratio of the effective prediction time length of the prediction model in unit time. The invention improves the power grid daemon stability of the compressed air energy storage system.

Inventors

XU MING
WEI WEI
WANG CUIPING
LI BOWEN
WU JIAHUA

Assignees

国华(诸城)风力发电有限公司

Dates

Publication Date: 20260508
Application Date: 20251226

Claims (10)

1. The intelligent conservation method for the compressed air energy storage system power grid based on the big data is characterized by comprising the following steps of: collecting operation data in a power grid of a compressed air energy storage system, cleaning, denoising and extracting features of the operation data to obtain operation features, training an initial model according to the operation features to obtain a prediction model, and predicting the operation state of the power grid by using the prediction model to obtain a prediction result; When the predicted result is at risk, switching the compressed air energy storage system from a shutdown or normal standby state to a power grid daemon hot standby mode, wherein when the risk actually occurs, a loading instruction is sent to the compressed air energy storage system in the hot standby state; acquiring the effective rate of the operation data in a single period, and determining whether the grid daemon stability of the compressed air energy storage system meets the requirement or not based on the effective rate of the operation data in the single period; if the grid daemon stability is not in accordance with the requirement, determining whether the length of a sliding average window needs to be increased; If the length of the sliding average window does not need to be increased, acquiring the transmission delay time of the operation data to determine whether the suitability of the update of the prediction model meets the requirement; if the suitability of the update of the prediction model does not meet the requirement, determining whether a dynamic threshold value of the minimum training data amount needs to be reduced or not; and if the dynamic threshold value of the minimum training data amount does not need to be reduced, determining the offset of the update trigger time based on the effective prediction duration duty ratio of the prediction model in unit time.
2. The big data based compressed air energy storage system grid intelligent daemon method of claim 1, wherein determining whether the grid daemon stability of the compressed air energy storage system meets a requirement based on the effective rate of the operational data in the single cycle comprises: Comparing the effective rate of the operation data in a single period with a preset second effective rate; If the effective rate of the operation data in the single period is greater than or equal to the preset second effective rate, determining that the grid daemon stability of the compressed air energy storage system meets the requirement; And if the effective rate of the operation data in the single period is smaller than the preset second effective rate, determining that the grid daemon stability of the compressed air energy storage system is not in accordance with the requirement.
3. The big data based compressed air energy storage system grid intelligent daemon of claim 2, wherein determining whether an increase in a moving average window length is required comprises: Comparing the effective rate of the operation data in the single period with a preset first effective rate and a preset second effective rate respectively; If the effective rate of the operation data in the single period is smaller than or equal to the preset first effective rate, determining that the length of the sliding average window needs to be increased; if the effective rate of the operation data in the single period is greater than the preset first effective rate and less than the preset second effective rate, the sliding average window length is determined not to need to be increased.
4. The intelligent conservation method of the compressed air energy storage system power grid based on big data according to claim 3, wherein the increasing amplitude of the sliding average window length is determined by presetting a difference between the first effective rate and the effective rate of the operation data in a single period.
5. The big data based compressed air energy storage system grid intelligent daemon of claim 4, wherein determining whether suitability of a predictive model update is satisfactory based on a transmission delay time of operational data comprises: Comparing the transmission delay time length of the operation data with a preset first delay time length; If the transmission delay time of the operation data is smaller than or equal to the preset first delay time, determining that the suitability of the update of the prediction model meets the requirement, and determining whether the length of the sliding average window meets the requirement; If the transmission delay time length of the operation data is longer than the preset first delay time length, determining that the suitability of the update of the prediction model is not in accordance with the requirement.
6. The big data based compressed air energy storage system grid intelligent daemon of claim 5, wherein determining whether a dynamic threshold that reduces the minimum amount of training data is needed comprises: comparing the transmission delay time length of the operation data with the preset first delay time length and the preset second delay time length respectively; If the transmission delay time of the operation data is longer than the preset first delay time and is smaller than or equal to the preset second delay time, determining a dynamic threshold value for reducing the minimum training data quantity; And if the transmission delay time length of the operation data is longer than the preset second delay time length, determining that the dynamic threshold value of the minimum training data quantity does not need to be increased.
7. The big data based compressed air energy storage system grid intelligent daemon of claim 6, wherein the magnitude of the decrease in the dynamic threshold of the minimum training data amount is determined by a difference between a transmission delay time of the operational data and a preset first delay time.
8. The big data based compressed air energy storage system grid intelligent daemon of claim 7, wherein determining the offset of the update trigger time based on the effective prediction duration duty cycle of the prediction model per unit time comprises: Comparing the effective predicted duration duty ratio of the prediction model in unit time with a preset duty ratio; If the effective prediction duration duty ratio of the prediction model in unit time is greater than or equal to the preset duty ratio, determining that the scene suitability of the prediction model update meets the requirement, and determining whether the dynamic threshold of the minimum training data amount meets the requirement without increasing the offset of the update trigger time; if the effective prediction duration duty ratio of the prediction model in the unit time is smaller than the preset duty ratio, determining that the scene suitability of the prediction model update is not in accordance with the requirement, and increasing the offset of the update trigger time.
9. The intelligent daemon of the power grid of the compressed air energy storage system based on big data according to claim 8, wherein the effective prediction duration of the prediction model in unit time is a ratio of a duration of a normal output prediction result of the prediction model in unit time to a total prediction duration.
10. The intelligent daemon of a compressed air energy storage system grid based on big data according to claim 9, wherein the increasing amplitude of the offset of the update trigger time is determined by the difference between the preset duty cycle and the effective prediction duration duty cycle of the prediction model in unit time.

Description

Intelligent conservation method for compressed air energy storage system power grid based on big data Technical Field The invention relates to the technical field of compressed air energy storage, in particular to an intelligent conservation method of a power grid of a compressed air energy storage system based on big data. Background The existing CAES system has a single operation mode, peak clipping and valley filling are usually carried out only according to electricity price signals, and the powerful synchronous generator and the quick starting capability of the existing CAES system are not used for supporting the stability of a power grid. The high-proportion new energy power grid has frequent voltage fluctuation and needs flexible reactive compensation means, and meanwhile, the power fluctuation is high in predictability, but the traditional unit has insufficient response. The thermodynamic process of CAES itself has inertia and cannot respond as much as a battery in milliseconds. But power changes in the grid (such as cloud cover movement, evening load climbing) tend to be predictable. The prior art fails to combine prediction with preparation to overcome the response delay of CAES. The Chinese patent publication No. CN119965833A discloses a response time optimization method of a compressed air energy storage power station based on time series prediction, which comprises the following steps of S1, collecting local historical environment and power generation and consumption data, predicting the current power generation and consumption data, S2, respectively establishing a local total power generation model and a local power consumption model by utilizing a GluonTS algorithm according to the data collected in the step S1, S3, establishing a scheduling instruction time sequence prediction model of a power plant according to the data of the local total power generation model and the local power consumption model in the step S2 and the historical data of a local power grid scheduling of the power storage power plant, S4, predicting the possibility of the power storage power plant being scheduled by the power grid in the future by utilizing the scheduling instruction time sequence prediction model of the power plant in the step S3, and S5, if the possibility of the scheduling instruction time sequence prediction model of the power plant exceeds a set threshold, namely the power grid scheduling is considered to be possible in the future, entering into preparation for the energy storage and power generation system so as to respond immediately when a scheduling instruction is received. Therefore, according to the time series prediction-based response time optimization method for the compressed air energy storage power station, whether the response time exceeds a fixed threshold value is taken as a basis for starting preparation, real-time fluctuation of transmission delay, data noise and the like of power grid operation data is not considered, when sudden disturbance occurs to the power grid, the fixed threshold value cannot be dynamically adjusted, and the problems that energy consumption is unnecessarily increased and the risk of untimely response causes insufficient protection stability of the power grid of the compressed air energy storage system are caused. Disclosure of Invention Therefore, the invention provides an intelligent daemon method for a power grid of a compressed air energy storage system based on big data, which is used for solving the problems that in the prior art, whether the fixed threshold is exceeded or not is taken as a basis for starting preparation, real-time fluctuation of transmission delay, data noise and the like of power grid operation data is not considered, when sudden disturbance occurs in the power grid, the fixed threshold cannot be dynamically adjusted, unnecessary energy consumption increase is caused, and the problem that the insufficient stability of the power grid daemon of the compressed air energy storage system is caused due to untimely response risk is caused. In order to achieve the above purpose, the invention provides a compressed air energy storage system power grid intelligent guarding method based on big data, comprising the following steps: collecting operation data in a power grid of a compressed air energy storage system, cleaning, denoising and extracting features of the operation data to obtain operation features, training an initial model according to the operation features to obtain a prediction model, and predicting the operation state of the power grid by using the prediction model to obtain a prediction result; When the predicted result is at risk, switching the compressed air energy storage system from a shutdown or normal standby state to a power grid daemon hot standby mode, wherein when the risk actually occurs, a loading instruction is sent to the compressed air energy storage system in the hot standby state; acquiring the effective rate of th