CN-121984123-A - Method, apparatus, device and medium for managing electricity consumption in novel electric load

CN121984123ACN 121984123 ACN121984123 ACN 121984123ACN-121984123-A

Abstract

The invention provides a power consumption management method, a device, equipment and a medium in a novel power load, which comprise the steps of obtaining current running state parameters of all novel power load assets containing electrochemical energy storage units or moving mechanical parts, calculating a superposition value of a circulating attenuation component and a calendar attenuation component as a health state attenuation quantity aiming at any novel power load asset based on the current running state parameters and candidate power scheduling instructions of the novel power load asset, converting the health state attenuation quantity into marginal attenuation cost based on the reset cost of the novel power load asset, the difference value of the current health state and the preset life-end health state and combining a user health preference coefficient, constructing a net benefit objective function by taking the power scheduling instructions of all novel power load assets as decision variables, and carrying out optimization solving on the net benefit objective function under the constraint of grid electricity price and auxiliary service demand signals to obtain an optimal asset power scheduling plan.

Inventors

ZHOU XIAOJIE
ZHONG JIANHUA
LIU JINGFENG
LI XIAOXIU
LING PENGYU
WEI HUI
LI RONGTAI
ZHU HUA
QIU JUNWEN
LUO MEIFANG
CHEN ZHENJIANG

Assignees

国网福建省电力有限公司上杭县供电公司
国网福建省电力有限公司
国网福建省电力有限公司龙岩供电公司

Dates

Publication Date: 20260505
Application Date: 20260120

Claims (10)

1. A method of power management in a new power load, comprising: acquiring current operation state parameters of each novel electric load asset containing an electrochemical energy storage unit or a moving mechanical component, wherein the operation state parameters at least comprise state of charge, current and temperature; Calculating, for any of the new power load assets, a superposition value of a cyclic attenuation component and a calendar attenuation component as a health state attenuation amount based on current running state parameters and candidate power scheduling instructions thereof, wherein the cyclic attenuation component is related to current, state of charge and temperature, and the calendar attenuation component is related to state of charge and temperature; Based on the reset cost of the novel power load asset, the difference value between the current health state and the preset life end health state, and combining the user health preference coefficient, converting the health state attenuation amount into marginal attenuation cost; and constructing a net benefit objective function containing the sum of a power grid service benefit item, a power consumption cost item and all marginal attenuation costs by taking power scheduling instructions of all novel power load assets as decision variables, and carrying out optimization solution on the net benefit objective function under the constraints of power grid price and auxiliary service demand signals to obtain an optimal asset power scheduling plan.
2. A method of power management in a new electrical load according to claim 1, wherein: the health state attenuation is calculated by using a personalized health attenuation model, and the personalized health attenuation model is constructed by the following steps: Acquiring static attribute data and historical operating state data of a novel power load asset, wherein the static attribute data comprises rated capacity, reset cost and battery chemical system type, and the historical operating state data comprises current, voltage, state of charge and temperature; According to the battery chemical system type, matching a corresponding basic electrochemical health decay mechanism model frame, wherein the model frame comprises a cyclic decay rate coefficient function and a calendar decay rate coefficient function which are provided with parameter sets to be calibrated, and the cyclic decay rate coefficient function and the calendar decay rate coefficient function are used for taking an operating state as a variable; based on the historical running state data, estimating and obtaining the real health state variable quantity of the novel power load asset through a data driving method; the error of the predicted health state variable quantity and the real health state variable quantity is minimized as a target, and the calibration value of the parameter set to be calibrated is obtained through optimization solution; and constructing and obtaining the personalized health attenuation model based on the calibration value and a basic electrochemical health attenuation mechanism model frame.
3. The method for power management in a new power load as set forth in claim 1, wherein the marginal decay cost is calculated by multiplying a user health preference coefficient by a health value loss, the health value loss being a product of a new power load asset reset cost and a health state decay amount divided by a difference between a current health state and a preset end-of-life health state.
4. The method for power management in a novel power load according to claim 1, wherein the attenuation rate of the cyclic attenuation component is positively correlated with the absolute value of the current and is regulated by the state of charge and the temperature, and the attenuation rate of the calendar attenuation component is accelerated with the rise of the state of charge and is influenced by the temperature to follow Arrhenius law.
5. A method of power management in a new electrical load according to claim 1, wherein: The candidate power schedule instruction is generated by: Determining the power adjustment range of each novel power load asset by combining an auxiliary service demand signal issued by a power grid, the current power grid load level and historical scheduling data; discretizing the power adjustment range into a plurality of power classes according to rated power and upper and lower limits of current available power of the novel power load asset, wherein each power class corresponds to duration matched with subdivision time periods in an optimization period; by combining different power levels with durations, a set of candidate power scheduling instructions is formed that includes different intensity schemes for conservative operation, medium load, and full load.
6. The method of claim 1, wherein the optimization model further comprises self-operating constraints of the new electrical load asset, including power constraints, energy storage capacity constraints, and temperature-controlled load comfort constraints.
7. A method of power management in a utility model as set forth in claim 1, wherein the power cost term is a product of grid electricity price and a total load comprising a sum of all utility model power load assets' loads and customer base line load predictions corresponding to auxiliary service demand signals, the customer base line load predictions being predicted from historical load data and a machine learning model.
8. The power consumption management method in a novel power load according to claim 2, wherein the historical operating state data is subjected to pretreatment steps of outlier rejection, time sequence alignment and missing value completion before being used for model construction, and the data driving method estimates the real health state change quantity by extracting key characteristic parameters of a historical charging curve and combining a pre-trained Gaussian process regression model.
9. An electrical power management system within a new electrical power load, comprising: A data acquisition module configured to perform the step of acquiring current operating state parameters of each new electrical load asset comprising an electrochemical energy storage unit or a moving mechanical component, the operating state parameters comprising at least state of charge, current and temperature, as claimed in claim 1; The attenuation quantification module is configured to execute the steps of calculating the health state attenuation amount and converting the health state attenuation amount into marginal attenuation cost for any new power load asset according to the claim 1, namely, calculating the superposition value of the cyclic attenuation component and the calendar attenuation component as the health state attenuation amount based on the current running state parameter of the asset and the candidate power scheduling instruction, and converting the health state attenuation amount into marginal attenuation cost based on the reset cost, the current health state, the preset life end health state and the user health preference coefficient of the asset; The optimizing and scheduling module is configured to execute the steps of constructing a net benefit objective function and optimizing and solving the net benefit objective function according to claim 1, namely, taking power scheduling instructions of all novel power load assets as decision variables, constructing the net benefit objective function comprising the sum of a power grid service benefit item, an electricity consumption cost item and all marginal attenuation costs, and optimizing and solving the net benefit objective function under the power grid price constraint, the auxiliary service demand signal constraint and the asset self-running constraint to obtain an optimal asset power scheduling plan.
10. A computer readable storage medium, on which a computer program is stored, characterized in that the computer program, when being executed by a processor, implements the method of any one of claims 1 to 8.

Description

Method, apparatus, device and medium for managing electricity consumption in novel electric load Technical Field The invention relates to the technical field of energy management, in particular to a power consumption management method, a device, equipment and a medium in a novel power load. Background With the promotion of the construction process of a novel power system taking new energy as a main body, virtual Power Plant (VPP) and load aggregator technology are taken as important carriers for realizing the interaction of source network and load storage, and are becoming key means for improving the flexibility and stability of a power grid. The method is characterized in that mass distributed resources such as electric automobiles, household energy storage systems, intelligent household appliances and the like are aggregated and cooperatively controlled through an advanced information communication technology to form a flexible-regulation negative power plant, so that auxiliary services such as frequency modulation, peak shaving, standby and the like of a power grid are deeply participated. This mode can effectively challenge the large-scale access of distributed power sources and exploit the regulatory potential of demand-side resources. In the conventional technology path, in order to achieve efficient centralized optimization scheduling, highly simplified modeling is typically performed on aggregated heterogeneous distributed resources. Specifically, these resources are abstracted as "ideal energy blocks" having only basic parameters such as upper and lower limits of charge and discharge power, energy storage capacity constraints, and charge and discharge efficiency. The core targets of the scheduling model are often set to maximize the market benefit of auxiliary services and minimize the electricity purchase cost of the external power grid, and meanwhile, the power grid tide safety, the node voltage limit, the user comfort level interval and the like are taken as constraint conditions. Within this framework, each scheduling instruction is essentially considered an indiscriminate, reversible energy throughput process for the controlled asset once, which simplifies while reducing model complexity, facilitates real-time decisions, but masks the physical essence inside the resource. However, the above conventional method has a significant limitation in that a large number of devices including electrochemical energy storage units (e.g., batteries) or moving mechanical parts are omitted, and physical and chemical changes during operation are not ideally reversible. Each charge-discharge cycle, rapid step change of power and frequent start-stop operation can cause irreversible abrasion, aging and other phenomena in the equipment, and the phenomena are directly expressed as accelerated decay of Health (SoH). This decay corresponds to a true asset worth compromising cost, whereas traditional models fail to take this important marginal cost into consideration scheduling decisions. This results in the system potentially overusing those assets that are otherwise aged or ill-conditioned, intangibly accelerating their end of life, and in a full life cycle perspective, rather reducing overall economics. Meanwhile, because flexible resources with lower health attenuation cost cannot be identified and preferentially called, the overall economy and operation sustainability of the aggregate resource pool are weakened, and personalized management requirements of users on own asset life are difficult to meet. Successful operation of virtual power plants is highly dependent on accurate state awareness, flexible resource aggregation, and optimized decision control techniques, which suggests that in-depth consideration of the internal state of the resources is critical. Disclosure of Invention In view of the drawbacks and deficiencies of the prior art, the present invention provides a method, apparatus, and computer-readable storage medium for power management in a novel power load. The method focuses on novel electric load assets containing electrochemical energy storage units or moving mechanical parts, obtains current running state parameters of each asset containing state of charge, current and temperature based on an optimization period, and converts the instructions into time sequence tracks reflecting the running state evolution of the asset through physical characteristics and energy balance principles by combining candidate power scheduling instructions. Based on an individualized health attenuation model constructed based on electrochemical mechanism and data driving fusion, two core modes of cyclic attenuation and calendar attenuation are integrated to calculate the attenuation quantity of the health state, the model determines a basic frame by matching the type of an asset battery chemical system, and parameter calibration is completed by combining historical operation data, so that the accuracy of attenuation q