Improving the Structure of the Electricity Demand Response Aggregator Based on Holonic Approach

Abstract

A demand response (DR) aggregator is a specialized entity designed to collaborate with electricity producers, facilitating the exchange of energy for numerous stakeholders. This application is a pivotal development within the Russian Energy System as it transitions to a Smart Grid. Its successful operation relies on the advancement and implementation of more efficient strategies to manage emerging energy assets and structures. The holonic approach is a distributed management model used to handle systems characterized by random and dynamic changes. This paper analyzes the specific aspects of the electricity demand management mechanism in Russia, primarily aimed at reducing peak load in the energy system by engaging active consumers who are outside the wholesale market. The DR-Aggregator is considered both a cyber-physical system (CPS) with a cluster structure and a business process. The DR-Aggregator exhibits essential holonic properties, enabling the application of a holonic approach to enhance the efficiency of the DR-Aggregator mechanism. This approach will facilitate greater flexibility in managing the load schedules of individual holon consumers, bolster the reliability of power supply by aligning load schedules among holon consumers within the super-holon cluster, and improve the fault tolerance of the DR-Aggregator structure, providing greater adaptability of demand management services.

1. Introduction

Establishing intelligent energy systems (IES) requires specialized approaches to ensure effective management of their structures and facilities under dynamic conditions. One of these approaches is the demand response (DR) mechanism, when, instead of building new generation sources, a redistribution of electricity consumption during peak hours and the supply of surplus power are initiated [1]. At the same time, consumers can change their consumption patterns and indirectly influence the electricity market. In addition, some DR programs are designed to provide these services quickly, which helps to improve the dynamic characteristics of the power system. DR programs decrease the overall load on the system, which helps to partially unload the system and provide a margin of safety in the event of system failures. Therefore, it is widely accepted that the DR mechanism can improve the reliability of the system by boosting both service availability and security. The DR mechanism can also increase the reliability of the power system by leveraging distributed generation sources [2]. To this end, the impact of the load level change caused by the DR mechanism is assessed, confirming that the peak load-shifting strategy can effectively improve the system’s reliability.

In [3], the DR mechanism is applied with the help of smart meters and IoT devices within the framework of a new load restoration method in the cyber-physical distribution system (CPDS). Under emergency conditions, the power supply topology is reconfigured using alternative sources, and the demand management process is launched to restore the critical load by disconnecting the non-critical one.

Increasingly sophisticated forms of DR mechanisms are introduced all over the world. Various pricing models for the DR mechanism operation are analyzed in [4]. This analysis shows that the choice of model type (setting different tariffs during the day, reducing critical peak load, real-time tariff changes, and prompt response to wholesale market price signals) affects the consumer participation degree in DR programs.

Several business models for DR mechanisms to increase energy flexibility are examined in [5]. This study indicates that the implementation of the DR mechanism usually addresses the integration of different renewable energy sources and their effects on the distribution network. However, it is necessary to include technical and socio-economic aspects, as well as their interrelations. It is also necessary to focus on the technical implementation of DR and the achievement of economic values, which lays the foundation for a demand response business model (DRBM).

In [6], the DR mechanism relies on the data on load from IoT devices. The DR program works successfully when the DR servers can transmit information about supply and demand in a timely and lossless manner. To improve data security, ref. [6] proposes using cloud computing as a serious breakthrough in managing the DR network. This study considers the simultaneous calculation of the energy and data balance in the DR network and takes into account the need to maintain power balance, secure network traffic, and service quality. Coordination of the consumers’ load flexibility for trading in the electricity markets in exchange for a certain percentage of revenue is carried out by a special structure called the Demand Response Aggregator (DR-Aggregator).

Specifics of the DR-Mechanism in Russia

A pilot DR mechanism project was initially launched in 2019 and continued until 2023. In early 2024, Demand Response was officially adopted as a system service.

DR-Aggregator is a coordinating center in the DR mechanism, designed to bring together the IES entities that have surplus energy resources for sale to their energy system yet do not participate in the wholesale electricity market [1]. The establishment of DR-Aggregators in Russia offers a promising solution to the problem of insufficiently managed resources to cover peak loads. The aggregator has the capability to engage consumers with generating capacities of up to 25 MW. These include local power systems that rely on small-scale generation with technological connections to the networks of regional grid companies.

Today’s electricity consumer is an active participant in energy consumption, storage, and generation based on innovative distributed generation technologies, demand response, etc. In the current landscape of transition to intelligent energy systems, every consumer—whether a power plant, an industrial facility, a hotel, an apartment complex, or others—functions as a cyber-physical system (CPS) that integrates information management with process subsystems, each influencing the other. We also consider the DR-Aggregator from the perspective of the CPS. As is known, the failure of one CPS subsystem (cyber or physical) can lead to the failure of the other subsystem (physical or cyber) and results in combined emergency conditions and even a failure of the entire structure.

A well-organized network-centric management system (NCM) contributes to the survivability of the DR-Aggregator [7]. In contrast to the traditional classical hierarchical structure inherent in any power system, there are horizontal connections at each level of the hierarchy in the NCM. Components of the same level exchange information with each other, which enhances the efficiency of decision-making and boosts the reliability of the DR-Aggregator structure. All levels of the NCM structure are both managing and manageable (except for the first level), and only the lowest technological level remains fully manageable as it houses the resources of active consumers. This structure is suitable for enterprises that are bound by contractual obligations with their aggregator cluster. If a DR-mechanism participant fails to meet its obligations during a period of low energy consumption and faces a shortfall of resources, it can seek support from neighboring active consumers that have surplus resources, as indicated by the red bidirectional arrows in Figure 1.

Modern smart technologies allow components at the lower level of the hierarchy to be involved in responding to changing conditions. Cyber-physical management (CPM) provides this opportunity to the lowest level [7]. The implementation of CPM becomes possible through the combination of CPSs and NCM when the participants of the DR-Aggregator are considered cyber-physical systems and can interact directly with each other.

In [8], the DR-Aggregator is considered as a business process. The required level of the business process detail, starting with a clearly defined task, helps to enhance its efficiency. The following benefits are achieved: transparency of the entire technological process chain, business process decomposition into functions in order to optimize operations and the speed of processing materials and data, the informed selection of the control structure with a fast response to emerging disturbances, and an increase in the speed of decision-making on process control.

In Figure 1, the black arrows of different thicknesses show market requests to the SO UES, the SO UES requests to the DR-Aggregator, and corresponding responses to the requests; the beige arrows indicate consumers’ remunerations for the energy they supply during the implementation of the DR-mechanism; the dotted lines depict the network-centric (NC) structure of the aggregator; the red arrows show the mutual assistance of customers.

The transformation of energy systems coupled with the advancement of information technologies allow for the use of new energy sources and their integration into the energy supply strategies and the demand response mechanism. Therefore, new ways to interact with energy facilities are required to solve the problems set more effectively. One of the methods is based on the holonic systems designed to simplify setting up and managing production processes, enabling power facilities to quickly respond to dynamic changes in the system, in particular, to market demands in the electric power industry [9].

This paper proposes using a holonic approach to improve the efficiency of the DR-Aggregator mechanism, relying on the previous developments of the authors, where the DR-Aggregator is considered as a CPS with a cluster structure using network-centric and cyber-physical management principles [7]. The holonic architecture takes into account user autonomy and implements recursive aggregation of users (as a system of systems) and dynamic reconfiguration of users with stochastic changes in the DR-Aggregator structure and its operating conditions. This approach enhances the stability and security of management and provides flexibility for demand management services, which is essential to the implementation of an intelligent network.

This paper has the following structure. Section 2 describes a holonic system in general with a specific focus on its application in the energy sector. Section 3 shows the approach proposed by the authors to improve the mechanism of the DR-Aggregator based on the capabilities of holonic systems. Section 4 demonstrates the algorithms available for use in a new landscape. The Discussion section addresses possible avenues to develop the topic presented in this paper.

2. Holonic Systems

2.1. Holonic Systems: The Emergence and First Application to Production (Holonic Manufacturing System)

The concept of holon [10] originated from the study and understanding of processes in biological and social systems: any system grows out of individual stable parts and itself becomes an independent stable form. Initially, holarchy was introduced into production [11]: holonic production is a distributed management paradigm aimed at solving problems with frequent changes and failures. Holonic production system (HPS) defines three types of basic holons: resource; product; and order.

• Holon resource consists of a physical part and an information part that manages the resource;
• Holon product includes functional characteristics (design, processes, product quality) and information about the technological process and products;
• Holon order is a production task: timing and management (dispatching) [12].

Each holon can include interacting subholons; at the same time, each holon may become a part of a superholon. The holonic approach is a hybrid of distributed and centralized approaches: autonomous subsystems adapt to a static structure, while the behavior of subsystems depends on the commands of the supervisory controller [13]. At each level (holons, subholons), they cooperate and compete, creating new properties in the system that are not inherent to each of them individually. This structure resembles both a hierarchy and a multi-agent system (MAS). However, the holonic approach implies dynamism and flexibility, and this is one of the differences from the traditional MAS approach.

The holonic systems approach is well suited for managing energy systems that are traditionally hierarchical, and their components (for example, microgrids) are part of a larger network.

2.2. Cyber-Physical Holonic System

Exhibiting the properties of self-development, self-organization, adaptation–biological production systems (BMS), reconfigurable manufacturing systems (RMS), holonic manufacturing systems (HMS), and other significant phenomena became the precursors of cyber-physical systems (CPS) [14]. In the CPS, sensors embedded in physical objects are responsible for collecting data in real time and transmitting data to the computing part of the system through a network/cloud environment (which provides decentralized management). This whole process is performed dynamically using feedback loops from the physical part to the cybernetic and vice versa.

In [15], it is indicated that the holon has the same function as the CPS and has two sides representing both digital (cyber) and physical attributes of the object (for example, devices). The holon structure consists of two main parts: informational (mandatory, responsible for high-level decision-making) and physical (if the holon is physical equipment and is responsible for transmitting decisions and control actions to the associated physical component). Thus, holon is defined as the CPS. Unlike holons, there is no such separation in the agent’s structure between the physical part and the information processing part.

2.3. Holonic Systems in the Energy Sector

As indicated in [9], the first mention of holons associated with energy systems was made in [16] in 2007. The Internet and energy markets were analyzed from the point of view of complex systems with some common behavioral and structural properties. A holonic smart grid (SG) is proposed in [17], where each user has sufficient autonomy to manage their own resources. The holonic SG structure is recursive, in which the user consists of smaller users at the lower level of aggregation and, at the same time, is part of a larger user at the higher level of aggregation. The holon consumers are recursively grouped into clusters to eventually form the holarchy of an intelligent network.

In [18], the concept of “holonic approach” is applied to the construction of a Smart Grid with the following triad {Holon-IES; Holon-IT; Holon communication system}; for Holon-IES at the following level, a “Generating consumer-holon” (GC-holon) is proposed—this is a consumer who independently produces electricity using Renewable energy sources. Its surpluses are stored in the energy storage system (ESS), which is able to exchange electricity with each other and with the main power system. In SG holarchy, the holonic properties of all components are manifested, and this contributes to increasing the energy efficiency of the Smart Grid:

• the autonomy of GC-holons provides them with flexibility in managing their load schedules;
• aggregation of GC-holons into super-holons can increase the reliability of the power supply and align load schedules between holons;
• dynamic reconfiguration of the structure increases fault tolerance.

In [9], the holonic concept is proposed for large-scale infrastructures, for example, IES, in order to make them more sustainable in real time and at different planning stages. Various IES facilities are structurally similar: sections of the power grid with different voltage levels connected to generators, loads, and ESS provide adjustable flexibility. Consumers are examples of holons who solve their individual tasks, taking into account market prices, technical limitations, and weather conditions.

Since future smart grids will bring together a large number of heterogeneous producers and consumers who are unpredictable and operate at different scales, ref. [19] presents a holarchy for intelligent network management with recursive integration of heterogeneous solutions that achieve contradictory goals. With the help of holarchy on various scales of intelligent networks (district, house), global tasks will be solved at the macro level, and local tasks at the micro level. In [20], a three-level holarchy of holons, “substation”, “feeder”, and “neighborhood”, was considered, where each holon was represented by an agent in a cyber-physical system.

Conceptually, the holonic energy system is represented by a set {Resource holon RH}, {Energy holon, EH}, {Service/Order holon, SH} [20] (Figure 2).

• resource-holon (RH) is a producer of electricity (consumer);
• energy-holon (EH) is the electricity produced (or product-holon, PH);
• service-holon (SH) describes how the IES provides a service to users (or order-holon, OH).

The RH–OH interconnections are used to organize the management of the energy system. The EH–OH connection allows you to evaluate the quality of user service.

2.4. Examples of the Holonic Approach Application to the DR-Aggregator Structure

In [17], the dynamic pricing model is applied to an architecture of holonic managing of subholons and the resource planner of holon consumers. Subholons are directed to the desired household load profile; the resource planner has to minimize the peak value of the total load of the energy community while ensuring that the resulting power flows comply with network restrictions.

Paper [21] presents the first attempt to develop a specific concept for the implementation of DADR, a decentralized active DR structure in which there is no connection between its individual elements (similar to the holonic structure). The initiation of load reduction depends on the frequency fluctuations measured and analyzed at the DADR installation point. DADR elements are supposed to be installed physically on the customer’s side, and one separate DADR element will be an interface between the distribution network and the devices being installed. The central feature of the proposed model developed to assess the impact of DADR technology on frequency control is a function describing the relationship between a change in frequency f(t) and a change in load power.

In this paper, it is proposed that holonic architecture be applied to improve the management principles of the DR-Aggregator in Russia, which is considered from the perspective of a cyber-physical system with a cluster structure. This will allow for flexibility in managing the load schedules of individual holon consumers, increase the reliability of power supply by aligning load schedules between holon consumers included in the superholon cluster, increase the fault tolerance of the DR-Aggregator structure using the possibilities of dynamic reconfiguration.

3. Applying a Holonic Approach to the DR-Aggregator

3.1. The DR-Aggregator as a Cyber-Physical Holonic System

In [22], the degree of detail of the holon interaction in the CPS is illustrated: a holon is a dual entity consisting of a physical part (Ph_holon) and a cyber part (C_holon), which, in turn, contains data about F_holon and an intelligent decision-making system for managing F_holon. The decision-making process is adapted to each level of holon detail in the global system. For example (see Figure 3), the holon consumer uses different equipment. The holon of the equipment consists of (C_equip, Ph_equip).

Each type of equipment is a set of smart components, for example, IED, consisting of (C_ied, Ph_ied). All (C_ied) exchange data among themselves and with (C_equip) to make internal decisions (decision-making level N). (C_equip) exchanges data with others (C_equip) and with (C_Consumer) for consumer decision-making (decision-making level N-1). Each (C_ Consumer) participates in making decisions corresponding to the degree of detail of its (Ph_ Consumer). This illustrates the property of recursiveness in holonic architecture.

Applying such a cyber-physical representation to the aggregator cluster structure, we see the exchange of information in the holarchy of everyone with everyone: at the lower level, smart devices pass their status to the functional block in the equipment; the equipment receives signals from the consumer and responds to them. Consumers exchange information both with the aggregator cluster and between themselves.

3.2. Examples of the Application of the Holonic Approach to the DR-Aggregator as to a Business Process

The author of [5] describes the components of the Business Model (BM), which, from an abstract point of view, are usually divided into three types (see

Table 1. General characteristics of the Business Model, Holarchy, and DR-Aggregator as a business process.

BM Components [5]	Holarchy	DR-Aggregator as a Business Process
Value proposition	Holon order	Market, System Operator, DR-Aggregator
Value creation and delivery	Holon resource	Aggregator cluster, Consumers–participatians of DR-Aggregator
Value capture	Holon product	Power energy, financial incentive

, first column). We have compared the BM, the holonic approach, as well as the DR-Aggregator as a business process. The results of the comparison are summarized in Table 1.

Such comparison shows that there is a reason to apply a holonic approach to the structure of the DR-Aggregator.

3.3. The Presence of Holonic Properties in the DR-Aggregator

Properties of holonic systems inherent in the DR-Aggregator:

• Recursiveness—DR-Aggregator (super-holon), aggregator clusters (holons), and consumers (sub-holons) are control systems with advanced ICT;
• Reconfiguration—the consumer can leave one aggregator cluster (for example, when the bandwidth of communication channels is limited) and connect to another one;
• Autonomy, cooperation, functional decomposition—consumers can be completely different in their type, composition, and goals, but it is easy for them to integrate into the structure of the DR-Aggregator and participate in peak-load reduction sessions in the power system;
• Flexibility—coordination of the reconfiguration process and the operation of control algorithms (in this case, recalculation following reconfiguration, which will be discussed below).

Each active consumer includes an adjustable or unregulated load. If the load is unregulated, then the active consumer must necessarily include ESS and/or its own generating capacities (usually RES); for an AC with a regulated load, these devices are not required. The holonic approach is the online interaction of active consumers who independently participate in reducing their consumption without going beyond technological limitations. What kind of load can be reduced? In fact, the holon resource prepares its business plan in advance:

• revises and corrects the network schedule of technological processes;
• performs a shift in the operating time of electric receivers with variable load;
• changes the operating mode of household electric receivers with flexible power control;
• uses individual generating capacities and ESS.

The lower-level consumers and second-level aggregator clusters are holons, and the upper-level DR-Aggregator is a super-holon. Thus, the hierarchical structure of the DR-Aggregator can be considered as a holonic cyber-physical system in which holons of various hierarchy levels are interconnected by electrical and communication networks, the topology of which may not coincide.

The communication network plays an important role in reliable energy management. The top-level DR-Aggregator exchanges information with the SO UES and aggregator clusters; aggregator clusters exchange information with the DR-aggregator and (active) consumers belonging to the cluster; consumers exchange information with the aggregator cluster and among themselves. Most DR programs assume the presence of a two-way communication network between the aggregator cluster and its active consumers. At the lowest level, there may not be an electrical connection between consumers, especially those belonging to different clusters, but it is necessary to organize communication via information transmission channels (at least Wi-Fi).

4. The Proposed Approach: DR-Aggregator as the Holonic System—Example of a Session of the DR-Aggregator Mechanism

4.1. The Application of the Holonic Approach to the DR-Aggregator in Russia

For clarity of application, we superimpose Figure 2 of the holonic structure on Figure 1 of the DR-Aggregator structure (shown in the form of a business process), and as a result, we obtain Figure 4.

Figure 4 demonstrates that the application of the holonic approach to the hierarchical representation of the DR-Aggregator makes it possible to conceptually transform its structure by dividing all model components into three categories of holons. The holon order is represented by managing entities (market, SO EPS, aggregator). Holon resources are consumers who release electric power by shifting the consumption schedule, reducing losses, and storing excess energy. The fact of the released energy into the grid is represented by a holon product.

The interaction of consumers with each other in a network-centric structure is shown in Figure 1 through dotted circles. In Figure 4, such a circle is preserved; but in fact, such interaction is a property of the holonic structure. Thus, when forming a holarchy to the DR-Aggregator structure of the network-centric principles of structure management are not clearly manifested; they are by default embedded in the holarchy itself—the interaction of every holon with everyone.

The location of the cyber-physical management (CPM) is at the lowest level of the holarchy, where information is exchanged between smart equipment components and provides an instant protective response to unforeseen events in the system. The drives involved in CPM are electrical components that switch the states of devices in accordance with technological solutions after receiving optimization results.

The holonic architecture allows you to combine the positive aspects of these approaches, separate the structure of the DR-Aggregator from the management algorithms, rebuild the business process of the DR-Aggregator, and arrange all components into holon categories. The newly formed holonic structure of the DR-Aggregator shows the need for the creation of new algorithms for information exchange between holon consumers and helps to typify the requirements for connecting consumers to the structure.

4.2. Tasks Solved by the DR-Aggregator Using Holonic Approach

Based on the above studies, the following approach to creating a holonic DR management system is proposed. In [4], various criteria for optimizing the operation of the aggregator are considered, but in the Russian energy system, the most relevant criterium in the DR-Aggregator operation conditions is the maximum profit of each active consumer, who is provided the required reduction in electricity consumption (reduction in active power consumption by the value ΔP) in the centralized power supply network. Such profit is determined by an incentive (payment) from the aggregator cluster for reducing consumption minus the cost of implementing the above measures aimed at reducing electricity consumption.

The initial data for participation in the consumption reduction session are power measurements, telesignals for switching the equipment on/off, technological parameters of electrical receivers, and a database describing permissible modes, restrictions, switching circuits of electrical equipment, and connections with power sources.

The main task of the DR session is to balance supply and demand in the energy system by using peak load-reduction and load-planning methods. To solve it, steady-state calculation methods are used for obtaining balanced generation and consumption values in individual nodes and over the network as a whole, i.e., to determine the amount of the load to be adjusted, as well as the power received from ESS and renewable energy sources necessary to calculate the consumer costs for reducing the load.

As such a method, in this paper, it is proposed to use the state estimation (SE) procedure, which implements the test equation (TE) method developed by Melentiev Energy Systems Institute [23,24], etc.

Test equations are steady-state equations containing only measured variables $y$:

$$w_k(y)=0$$

(1)

Equation (1) is called the test equation because, by the magnitude of the discrepancy obtained after substituting the measurement values in it, one can judge the reliability of these measurements. When using the TE method, the SE task is to minimize the weighted least squares criterion:

$$J(y)={(\overline{y}-\widehat{y})}^T{R}_y^{-1}(\overline{y}-\widehat{y})$$

(2)

that is, it consists of calculating estimates $\widehat{y}$ of the measured variables $\overline{y}$ under restrictions in the form of TE system (1) and restrictions in the form of inequalities on the technological limits of the mode parameters (voltages in nodes, generation capacities, power flows along power lines). In Equation (2), there is the vector of measured parameters; $R_y$ is the covariance matrix of measurement errors.

To solve the objective function, a Lagrange function is compiled (where $\lambda -$is the vector of indeterminate Lagrange multipliers)

$$L=J(y)+\lambda w_k(y),$$

(3)

the minimization of which gives an expression for determining the estimates of the measured variables:

$${\widehat{y}}^{(i+1)}={\widehat{y}}^{(i)}-R{\left({\displaystyle \frac{\partial w_k}{\partial y}}\right)}^T{\left[\left({\displaystyle \frac{\partial w_k}{\partial y}}\right)R_y{\left({\displaystyle \frac{\partial w_k}{\partial y}}\right)}^T\right]}^{-1}w({\widehat{y}}^{(i)})$$

(4)

The problem is solved iteratively until the TE discrepancies $w_k(y)$ are less than an accuracy threshold.

The procedure of Bad Data Detection (BDD), mentioned later in the text, is based on this check [25].

The limitations in the form of inequalities are taken into account when making corrections calculated according to (4).

This method allows you to obtain a mode as close as possible to measurements, take into account the limitations in the form of equalities and inequalities, and set the estimates of the necessary variables equal to the measured values, setting zero variances to them to check initial information (to perform Bad Data Detection (BDD) procedure, which may be distorted due to technical failures in ICS or cyber attacks.

4.3. Distributed State Estimation (SE) with a Centralized and Holonic Control Structure of the DR-Aggregator

To calculate the modes of the IES, considered as a set of subsystems included in it, distributed SE technologies are used [26,27], etc.

Figure 5a,b shows the control system options for the three-level DR-Aggregator hierarchical structure, where the first level is the DR-Aggregator level; the second level consists of the aggregator clusters, and the third level is the active consumers included in the clusters.

The main difference between structures Figure 5a,b consists in hierarchical management: control commands are transmitted strictly from a higher control center to a lower one, and information concerning the state of objects (measurements, circuit topology, calculated values of mode parameters) and the implementation of control functions are transmitted from bottom of the hierarchy to its top (in opposite). At the control center level, it is assumed that the entire scheme is controlled based on full observability using measurements coming from each area of the corresponding hierarchy level. When implementing the holonic management of the DR-aggregator, it is assumed that each holon of a certain level of the hierarchy is connected to the holons of its level and the appropriate control center. It receives information from subordinate levels, autonomously coordinates it for its level, and provides it to a higher level.

With the centralized DR-Aggregator management, which is currently implemented, the distributed SE at the third level of the hierarchy assumes the calculation of estimates for each active consumer independently of each other (Figure 5a). To coordinate the calculations obtained, it is necessary to record the calculated values of the voltage magnitude and angle and the amount of power consumption reduction (Ui, δi, ΔPi) at the junction of the active consumer to the centralized power supply network or fulfill iterative calculations. If there are electrical connections between the participants in the cluster, then the values of voltage magnitudes and angles must be recorded at the nodes of these connections. The obtained estimates of the participants’ parameters are transmitted to the control centers of the aggregator clusters, where the SE procedure is performed for the entire scheme of each aggregator cluster, and the SE results are transmitted from the clusters to the Control Center of the DR-Aggregator. The calculation mode for the entire hierarchical structure of the DR-Aggregator is formed at the top level of the hierarchy. To avoid lengthy iterative calculations, PMUs are installed to fix the boundary variables in the boundary nodes, which record synchronized phasor measurements in the installation nodes with high accuracy. It should be noted that there is insufficient coverage of the lower level of the hierarchy by telecommutations, as well as the absence of PMUs; therefore, long iterative calculations are possible to coordinate the results of individual consumer’s operating systems.

For the holonic control of the DR-Aggregator structure, we implemented the Fully Distributed State Estimation procedure proposed in [28,29], but on the the TE method basis. As with centralized management, the calculation of estimates begins with a local SE at the lowest level of the hierarchy but with the involvement of data from related cluster participants (bi-directional arrows in Figure 5b) to form additional test equations. Edge-computing technology can be used to perform the local SE procedure [30] at the boundary close to the measuring devices; this ensures minimal delay in transmitting data to the control center and increases their confidentiality. Cloud technologies can be used to provide access to distributed data from neighboring clusters that are not connected by data transmission channels [31]. If necessary, this information can also be accessed through cloud storage for other objects included in the DR-Aggregator structure.

To implement this approach, compared with the hierarchical distributed SE discussed above, local observability of all control areas is no longer required. In addition, the proposed algorithm is more computationally efficient since it does not require iterative coordination of the local SE results.

Since active consumers operate in constantly changing conditions, then at certain time intervals, there may be an excessive generation or excessive load in some consumer holon, and such a regime cannot be balanced without violating technological restrictions, as evidenced by the results of the local condition assessment procedure. Such a holon can move from one DR cluster to another, as shown in Figure 5b (the right active consumer from cluster 1 switches to cluster 2, optimizing resource usage). When there is a potential risk of failure in the operation of its aggregator cluster, the consumer can go offline. This flexibility leads to a dynamic reconfiguration of the holarchy of the DR-aggregator over a period of time, which contributes to self-healing and increases the efficiency of its structure.

4.4. Example of a Session of the DR-Aggregator Mechanism

The market’s advance request to SO EPS and the same advance SO EPS request to the DR-Aggregator are expressed by the value of ΔP within the limited time interval [T1; T2].

At the moment T1, a DR-session starts; the holonic scheme of the DR session for one separate cluster (in the form of three holons) and the information flowing between them are shown in Figure 6.

The DR-Aggregator divides the value ΔP by ΔP 1, ΔP 2, …, ΔP n in accordance with the coefficients p1, p2, …, pn:

ΔP j = ΔP × pj,

where n is the number of aggregator clusters.

• Holon order implements the functions of the control subsystem of the aggregator cluster;
• Holon product is an information and communication subsystem of the aggregator cluster and its customers; this holon contains a “product model” in our case, which is the equivalent scheme for objects included in the cluster and the entire cluster. It receives data on the current state of objects (TI mode parameters and TS on the status of switching equipment);
• Holon resource is a physical subsystem of the aggregator cluster and its customers; it contains physical objects of the customers that are part of the cluster (consumers of electricity (loads), ESS, RES, etc.) and sensors that capture information about the topology and scheme parameters of this cluster and its customers.

Our investigation in applying a holonic approach to DR-Aggregator is in the very beginning; that is why, at this stage, we can only design a new holonic algorithm.

4.5. Algorithm of the DR-Mechanism Session as the Holonic System

• Holon order:

The jth aggregator cluster distributes the value of ΔP j received from the aggregator to ΔP j1, ΔP j2, …, ΔP jm in accordance with the coefficients l1, l2, …, lm,

ΔP ji = ΔP j × li,

where m is the number of participants that sends information to participating consumers.

• According to the unit commitment that belongs to holon resources, an analysis of its technical condition is performed, and a conclusion about the possibility of participation of holon resources in further DR sessions is made;
• In case of disconnection of a participant, the declared ΔP j value is adjusted, and the li coefficients of those participants who agreed in advance to increase the volume of released power in force majeure event are recalculated to ΔP j recalc. Update ΔP j recalc assign as a request;
• The order for the execution of the DR-session and the ΔP j values for each cluster are transmitted to the holon product;

• Holon product:

4.

Comparing the request and the order—did the FDI attack occur while receiving information on the order from the holon order? If the values of the request and the order do not match, inform all the holon resources and the aggregator cluster (holon order);

5.

Requesting data on the current state of cluster objects (remote measurements and telesignals on the state of switching equipment at the holon resource);

6.

Checking telesignals, what changes in the switching circuit occurred by the time T1, and the formation of the current calculation scheme. Running BDD procedure with TE method using measurements from neighbor resource holons (consumers);

7.

Starting SE: Calculate the internal balance of the participant and check whether the voltage level remains normal when the load is reduced. If not, then the ΔP value is reduced to a value that provides a technological voltage level. The adjusted ΔP corr value is transmitted to the holon order;

8.

Calculation of the released power by reducing the regulated load (ΔP n), using electricity from ESS ((ΔPess) and/or from RES (ΔPres):

ΔPcalc = ΔP n +ΔPess + ΔPres;

9.

Sending the calculated load reduction values to the physical subsystem of the aggregator cluster (to the holon resource);

10.

Holon resource: It checks whether each consumer can realize the calculated value of reducing consumption according to the technical condition of its components (Is it possible to shift the load to off-peak hours? Is there sufficient charging of the ESS? Is it possible to use electricity from renewable energy due to weather conditions? etc.). If not, then the calculated values of the ΔP are replaced by possible realizable values of the ΔPactual, which are transmitted to the information subsystem (in the Holon product at point 8);

11.

Initiation of the calculation of monetary compensation Di for the i-th consumer according to the formula

Di = ΔPi * di –( ΔPn × Cn +ΔPess × Cess + ΔPres × Cres),

where di is the specific remuneration of the i-th active consumer for reducing the consumption of active power (per unit); Cn, Cess, and Cres are the specific costs of reducing power consumption due to the regulated load, connection of ESS and RES;

12.

Calculation of monetary compensation for the jth aggregator cluster:

Dj = ΣDi.

The transfer of this value to the holon order.

5. Discussion

Since 2024, the DR mechanism has been a system service for the wholesale market. The ongoing transition from EPS to IES dictates the need to revise the traditional hierarchical structure adopted in power systems across all levels (system-wide, regional, local). Modern electricity consumers may be partially or completely independent of the central power supply system, but this does not prevent them from participating in the demand response mechanism. To ensure the effective functioning of this mechanism, it is essential to consider the seamless connection of all consumers wishing to participate in the DR session. This must be achieved while adhering to several key restrictions, including participation timing, the amount of available power, electricity quality, and strict fulfillment of obligations, among others.

The studies performed allow for the conclusion that the holonic system is the most effective approach for enhancing the efficiency of the DR mechanism. Each participant is a holon, i.e., an independent, self-sufficient, and cooperative entity. It has the ability to integrate with higher-ranked holons while simultaneously comprising lower-ranked ones. The recursive and self-similar nature of holons, along with their cyber-physical interpretation, creates universally applicable decision-making processes across various levels of the holonic structure. Coordinating holons to develop collaborative solutions greatly enhances their effectiveness. The structural decomposition of the system components allows for the separation of the system structure from the system control algorithm.

The representation of the DR-Aggregator as a business process enables the following structuring: service holon (the actual DR mechanism of the control system); resource holon (aggregator clusters and consumers included in them); and product holon (available power, electricity quality). Implementing the holonic approach not only updates the existing algorithms but also facilitates the development of new ones.

Section 4 demonstrates the application of algorithms utilized by authors working under new conditions.

In contrast to traditional distributed state estimation, which requires equal voltage magnitudes and angles at the boundary nodes of two neighboring regions or equal absolute power flow values at the ends of the boundary lines, the holonic approach allows for complete independence of all neighboring entities. Therefore, when managing each participant (holon resource) in the load reduction process, it is essential to maintain an internal power balance while considering voltage level constraints.

The absence of technological load reduction within the holon resource indicates that its load is powered by internal resources such as energy storage systems (ESS) and renewable energy sources (RES). In this case, it is essential to implement new algorithms that describe the processes of energy generation from solar and wind (possibly combined) farms, as well as the dynamics of ESS discharging and subsequent charging.

The aggregator cluster efficiently monitors the consumer statuses, ensuring that if one or several holon resources become disconnected during the demand response session, the participation coefficients of the remaining participants are immediately recalculated based on their assumed obligations.

6. Conclusions

In the context of the transformation of energy and the development of information technology, there is an opportunity for a new way of organizing the interaction of energy facilities to solve the tasks set more effectively. It is proposed to use a holonic approach to improve the principles of management in the operation of the DR-Aggregator. To implement the holonic approach, the DR-Aggregator is considered from the standpoint of the CPS and the business process, using network-centric and cyber-physical principles of managing the structure of the DR-Aggregator. The holonic architecture allows for combining the positive aspects of these approaches, separating the structure of the DR-Aggregator from the management algorithms, rebuilding the aggregator’s business process, and separating all components into holon categories. The newly formed holonic structure of the DR-Aggregator shows the demand for the creation of new algorithms for information exchange between consumers–holons and helps typify the requirements for connecting consumers to the structure.