CN-122000954-A - Industrial and commercial energy storage dynamic control method based on user electricity utilization behavior prediction

CN122000954ACN 122000954 ACN122000954 ACN 122000954ACN-122000954-A

Abstract

The invention discloses a power consumption behavior prediction-based industrial and commercial energy storage dynamic control method, which relates to the technical field of energy storage system control and comprises the following steps of S1, edge-cloud cooperative control system deployment, S2, multi-source data and characteristic acquisition, S3, load prediction model construction, S4, dynamic optimization control, S5, backflow prevention and demand control; the method realizes intelligent dynamic scheduling of the energy storage system by constructing an edge-cloud cooperative layered control architecture, data acquisition covers multidimensional characteristics such as load curves, equipment start-stop events, environmental parameters, photovoltaic output, electricity price information and the like, comprehensive input is provided for a prediction model, dynamic optimization control is based on the model prediction control framework, the purpose of minimizing comprehensive electricity consumption cost is achieved, photovoltaic absorption efficiency and electricity charge optimization effect are remarkably improved, and through cooperation of edge layer millisecond response and cloud intelligent decision, the problem of adjustment hysteresis caused by overlong period of a traditional control strategy is solved, and an economic and efficient energy storage system solution is provided for industrial and commercial users.

Inventors

SONG KEJUN
LIU QIN
HAO DONGWEI
Sang Chengkuan

Assignees

江苏领储宇能科技有限公司

Dates

Publication Date: 20260508
Application Date: 20251229

Claims (10)

1. A dynamic control method for industrial and commercial energy storage based on user electricity utilization behavior prediction is characterized by comprising the following steps: S1, deploying an edge-cloud cooperative control system, namely constructing a hierarchical control architecture and defining functions of each hierarchy; S2, multi-source data and characteristic acquisition, namely acquiring real-time data; s3, constructing a load prediction model by adopting Carrying out load prediction by the hybrid neural network; S4, dynamic optimization control, model-based predictive control Dynamically optimizing the energy storage charging and discharging power; s5, countercurrent prevention and demand control, namely implementing a comprehensive countercurrent prevention and demand control strategy according to the energy storage state, And dynamically adjusting the power of the grid-connected point to charge and discharge energy storage or photovoltaic output.
2. The method for dynamically controlling industrial and commercial energy storage based on the prediction of the electricity consumption behavior of a user according to claim 1, wherein the construction of the hierarchical control architecture defines the functions of each hierarchy, and the specific method comprises the following steps: The edge-cloud cooperative control system comprises an edge layer, a cloud layer and a communication framework; The edge layer adopts The server is used as edge computing equipment and is prepared from Arian cloud The cloud platform is used for data preprocessing, anti-backflow quick response and power gradual change control; The cloud layer is deployed in the Arian cloud A platform for Model prediction, Rolling optimization and life health assessment.
3. The method for dynamically controlling industrial and commercial energy storage based on user electricity behavior prediction according to claim 2, wherein the communication architecture is a data transmission channel connected with a device layer, an edge layer and a cloud layer; the device layer adopts Protocol, edge-cloud adoption Protocol, communication security pass-through Encryption and two-way certificate authentication guarantee.
4. The method for dynamically controlling industrial and commercial energy storage based on prediction of electricity utilization behavior of a user according to claim 1, wherein the real-time data comprises load data, equipment state, environmental parameters, photovoltaic data and electricity price information; The load data is collected through the intelligent ammeter, and the load power calculation formula is as follows Wherein Indicating the current point in time or instant in time, A window of sampling times is represented and, Representation of The instantaneous value of the voltage at the moment in time, Representation of Instantaneous value of current at time; the equipment state is that an equipment start-stop event is identified through the intelligent ammeter; The environmental parameters are used for acquiring temperature through a built-in sensor of a weather station or a photovoltaic inverter And an illumination intensity H; The photovoltaic data are communicated with and uploaded with photovoltaic power data through a photovoltaic inverter; the electricity price information passes through the power grid The interface acquires the time-of-use electricity price matrix in real time.
5. The method for dynamic control of energy storage in industry and commerce based on prediction of electricity consumption behavior of user according to claim 1, wherein the following steps are adopted The load prediction method of the hybrid neural network comprises the following specific steps: the said The hybrid neural network consists of an input layer, A layer(s), The input layer receives a feature matrix comprising a plurality of time steps and a plurality of feature dimensions; the layer comprises a first number of neuron units for extracting long-period trend features of the electrical load; The output layer comprises a plurality of prediction nodes and outputs a load prediction curve of a future preset time length; model training adopts a loss function, and the expression is as follows Wherein Is that Loss, used for the stable training, In order to enhance the loss at the required measuring point, Enhancing losses, coefficients for instantaneous plant start-stop characteristics And For adjusting the importance of different penalty terms; Training process deployment in alicloud The platform is configured to accelerate model convergence by adopting preset learning rate and batch size parameters, and the reasoning stage is deployed at the edge layer And the server uploads the real-time data to the cloud end according to a preset period, the cloud end rolls and updates the model based on the historical data of the preset time length, and the load prediction result of the preset time length in the future is sent to the edge layer to support dynamic control decisions.
6. The method for dynamic control of energy storage in industry and commerce based on prediction of electricity utilization behavior of users according to claim 1, wherein the control is based on model prediction The specific method for dynamically optimizing the energy storage charging and discharging power comprises the following steps: the dynamic optimization control is controlled by model prediction Implementation, performed by edge devices Optimizing strategy to dynamically adjust energy storage charge and discharge power, and monitoring anti-reflux state of edge layer with preset short period, wherein the anti-reflux state comprises photovoltaic power generation power Load power Energy storage system A state; the said Optimization is core with objective function minimization, and the expression of the objective function is Wherein For the time-step index, In order to predict the total number of steps in the time domain, For the weight coefficient of the electricity charge, For the weight coefficient of the time-sharing electric charge, As a coefficient of the gain of the photovoltaic, In order to be required for the electricity price, In order to realize the time-sharing electricity price, For the benefit of the photovoltaic power generation, For grid-tie point (check point) power, The power is photovoltaic power; the optimization process is limited by the following constraints: Wherein In order for the charging power to be high, For the power of the discharge it is, Is an energy storage converter The power rating of the device is set, In order to achieve the efficiency of the charge, In order for the discharge efficiency to be high, 、 To be a safe operating interval boundary for the stored state of charge, A preset proportion coefficient for charging power, A preset scaling factor of the discharge power.
7. The method for dynamically controlling industrial and commercial energy storage based on the prediction of the electricity consumption behavior of a user according to claim 1, wherein the implementation of the comprehensive anti-reflux and demand control strategy comprises the following specific steps: Obtaining photovoltaic power generation power in real time Energy storage real-time charging power Energy storage real-time discharge power Grid tie point power Load power Of energy storage systems A state; Real-time charging power based on energy storage Energy storage real-time discharge power Judging the charge and discharge states of the energy storage system, and combining the grid-connected point power Of energy storage systems And a state in which a demand control strategy and an anti-reflux control strategy are respectively executed.
8. The method for dynamically controlling industrial and commercial energy storage based on prediction of electricity consumption behavior of a user according to claim 7, wherein the specific method for judging the charge and discharge states of the energy storage system is as follows: Retrieving power thresholds from a database Real-time detection of energy storage real-time charging power through energy storage converter And energy storage real-time discharge power In combination with a power threshold Judging the current running state: When (when) When the energy storage system is in a charging state; When (when) In the time-course of which the first and second contact surfaces, judging that the energy storage system is in a discharge state; When (when) And is also provided with And when the energy storage system is in a standby state, judging.
9. The method for dynamically controlling industrial and commercial energy storage based on the prediction of the electricity consumption behavior of the user according to claim 7, wherein the executing the demand control strategy comprises the following specific steps: When the energy storage system is in a charged state, if it meets Setting the charging power Otherwise, the device operates according to the charging power issued by the cloud, wherein For the preset demand alert factor, The upper limit power is controlled for the demand.
10. The method for dynamically controlling industrial and commercial energy storage based on the prediction of the electricity consumption behavior of the user according to claim 7, wherein the executing of the anti-reflux control strategy comprises the following specific steps: When the energy storage system is in a discharging state or a standby state, if And is also provided with Then set up ; If it is And is also provided with Setting the photovoltaic power generation power ; If it is And is also provided with Then set up ; If it is And is also provided with The energy storage system operates according to a cloud strategy, wherein For the maximum safe operating boundary of the stored state of charge, In order to preset the anti-reflux adjustment coefficient, The upper limit power is controlled for the required amount, The power is rated for the energy storage converter.

Description

Industrial and commercial energy storage dynamic control method based on user electricity utilization behavior prediction Technical Field The invention relates to the technical field of energy storage system control, in particular to a machine learning prediction-based industrial and commercial energy storage dynamic scheduling method. Background Along with energy consumption upgrading and double-carbon target pushing in the industrial and commercial fields, the energy storage system is used as core equipment for optimizing energy configuration and improving energy utilization efficiency, and a dynamic control technology of the energy storage system is paid attention to. The current industrial and commercial energy storage system faces three technical bottlenecks in the dynamic control field, the economical efficiency and the energy utilization efficiency are severely restricted, the control strategy is rigidified and invalid, the traditional EMS system adopts the charging and discharging of a preset peak-valley period, the dynamic load mutation cannot be responded, for example, the actual measurement data of a certain Jiangsu manufacturing factory show that the maximum demand exceeds the standard rate by 27% under the charging and discharging strategy of a fixed period, the required electricity charge is only reduced by 9% -12% (which is far lower than the theoretical optimization space by 20% -25%), the control period is longer than 2min, and the quick load change cannot be responded timely. The load prediction precision is insufficient, the existing prediction model only depends on a macroscopic load curve and meteorological data, the prediction error is as high as 15% -22% due to the fact that equipment start-stop events are not associated, meanwhile, the association influence of real-time illumination intensity on photovoltaic output and load is generally ignored, and the prediction deviation of photovoltaic digestion is more than 25%. The traditional LSTM model cannot capture 30 ms-level instantaneous power change of equipment start and stop due to the fact that the time step is 15min, so that load mutation point prediction lag is caused. The traditional countercurrent prevention scheme mainly uses direct light rejection or photovoltaic output limitation, and cannot fully utilize the photovoltaic capacity, so that obvious benefit loss is caused. Disclosure of Invention The invention aims to provide a dynamic control method for industrial and commercial energy storage based on user electricity utilization behavior prediction, which solves the problems in the background technology. The technical scheme is that the industrial and commercial energy storage dynamic control method based on the user electricity utilization behavior prediction comprises the following steps of S1, deploying an edge-cloud cooperative control system, constructing a layered control framework, and defining functions of all layers. S2, multi-source data and characteristic acquisition, namely acquiring real-time data. S3, constructing a load prediction model by adoptingThe hybrid neural network performs load prediction. S4, dynamic optimization control, model-based predictive controlAnd dynamically optimizing the energy storage charging and discharging power. S5, countercurrent prevention and demand control, namely implementing a comprehensive countercurrent prevention and demand control strategy according to the energy storage state,And dynamically adjusting the power of the grid-connected point to charge and discharge energy storage or photovoltaic output. Preferably, the hierarchical control architecture is constructed to define functions of each hierarchy, and the specific method is that the edge-cloud cooperative control system comprises an edge layer, a cloud layer and a communication architecture. The edge layer adoptsThe server is used as edge computing equipment and is prepared from Arian cloudThe cloud platform is used for data preprocessing, anti-backflow quick response and power gradual change control. The cloud layer is deployed in the Arian cloudA platform forModel prediction,Rolling optimization and life health assessment. Preferably, the communication architecture is a data transmission channel connecting the device layer, the edge layer and the cloud layer. The device layer adoptsProtocol, edge-cloud adoptionProtocol, communication security pass-throughEncryption and two-way certificate authentication guarantee. Preferably, the real-time data includes load data, equipment status, environmental parameters, photovoltaic data, electricity rate information. The load data is collected through the intelligent ammeter, and the load power calculation formula is as followsWhereinIndicating the current point in time or instant in time,A window of sampling times is represented and,Representation ofThe instantaneous value of the voltage at the moment in time,Representation ofInstantaneous value of current at time. The equipment