RU2812195C1 - Method for intelligent load control in isolated power systems in emergency modes and device for its implementation - Google Patents

Abstract

FIELD: electrical engineering.

SUBSTANCE: device for intelligent load control in isolated power systems in emergency modes is made with the possibility of integration into the control system of electric power facilities and containing a set of hardware, a workstation, a set of software. The hardware set consists of an intelligent load control subsystem (ILCS), an load shedding executive system (LSES), and local load control devices (LLCD). ILCS is configured to interact with upper level control devices - software and hardware complex for intelligent voltage and reactive power control (IVRPC SHC) to minimize technological losses in the electrical network, and lower level - load shedding executive system (LSES). LSES is configured to receive control signals from emergency control automation, as well as to determine the list of disconnected loads and transmit a remote control signal for disconnecting loads to local load disconnection devices. LLCD are made with the ability to control the state, load value, power quality parameters of the electrical installation of the consumer of a separate connection, as well as the ability to turn off or turn on the load.

EFFECT: increased service life of isolated power systems in emergency modes, increased reliability of power supply to consumers, prediction of emergencies and preventing development of accidents.

7 cl, 3 dwg

Description

Technical field

The claimed invention relates to the field of electric power and is intended for use in the implementation of intelligent control of operating modes of electric power systems.

State of the art

Providing a normal mode of supply of electricity to consumers in an isolated energy system, without interruptions, except in cases of emergency electric power conditions (the threat of a shortage of electrical energy and power), or violation of the characteristics of the technological connection specified in the documents on the technological connection (including exceeding the maximum power consumer power receiving device).

In any power system, power consumption can reach a peak value at any given time. It is obvious that in the interests of ensuring economical and efficient energy consumption, widespread implementation of intelligent automated control systems is required. Their use on the medium and low voltage side is undoubtedly a significant factor in the modern concept of building a smart electrical network in order to increase the operational reliability, durability and efficiency of energy equipment, to solve the problems of dispatch, production, technological and organizational and economic management of the energy sector.

At the stage of transition to smart and flexible electrical networks (Smart Grid), an intelligent system is needed that will combine all the loads of the power system into a single controlled complex, which will streamline the consumption of electrical energy and eliminate the uncontrolled operation of consumers.

The following are known from the prior art:

An intelligent electricity meter containing a control unit connected to a memory unit and an indicator, contains a means of cryptographic information protection and an interface unit connected in series to the control unit; a network parameter measurement unit, a load limiting unit, a real-time clock, a sensor unit connected by a star to the control unit, wherein the cryptographic information protection means contains a real-time clock. An expansion of the arsenal of technical means is being achieved (RU 188731 U1, 2019.04.23).

A method for monitoring an electrical load having current and voltage generated by a power supply associated with the load, which includes the following steps: introducing a detection module between the load and the power supply to determine one or more characteristics of the load; combining the current or voltage calibration signal and the electrical signal into a combined signal; applying a mathematical transformation to the combined signal in the detection module; measuring one or more load characteristics using the combined signal; and determining operational parameters for determining one or more load characteristics. (RU 2486650 C2, 2013.06.27).

A method for monitoring and managing electrical energy consumption by consumers in a home includes setting priorities for consumers in terms of reliability and the method of their switching, transferring monitoring information to a microcontroller, and managing consumption. The complex that implements the method includes the electrical network of the house, made in the form of separate lines, means for measuring power, load control means, and disconnecting devices (RU 2725023, C1 2020.06.29).

Disclosure of the invention

The technical result consists in increasing the service life of isolated power systems and increasing the reliability of power supply to consumers in isolated, hard-to-reach and island territories.

The technical result is achieved by the fact that in order to increase the service life of isolated power systems and increase the reliability of power supply to consumers in isolated, hard-to-reach and island territories, a method of intelligent load control in isolated power systems in emergency modes is used, the essence of which is to eliminate unnecessary shutdowns of consumers' power receivers and minimize the consequences of emergency disturbances in isolated power systems, reducing the number of failures of main process equipment.

A method for intelligent load control in isolated power systems in emergency modes in which:

collect information about the current state of the power system, for which they use an information collection and transmission system (ICTS), which accumulates information about the state of the power system from the existing operational information complex of an isolated power system (OIC) by initiating the process of preparing measurements of modes suitable for modeling and optimization, such as :

- obtaining archived sections of teleinformation and network state, with the help of which machine learning and the initial launch of the forecasting process are performed;

- receiving and storing in the database sections of television information with a given frequency of current sections of television information and network status; logical control of the integrity of the received data using a set of control actions;

- formation of a data slice in the “nodes/branches” model, while linking the teleinformation to a single-line diagram of the electrical network, while the teleinformation slice is a set of telemeasurements and telesignals about the state of the network, taken for a certain time, and the simultaneity of time is determined by the time stamps present in each measurement, while the time stamp is transmitted from the final digital measuring device, while the teleinformation sections include:

- telemetry of active and reactive (P, Q) power along power lines, including lines going to the consumer;

- voltage measurements on station and substation buses; state of switching devices;

- operating mode of local control devices:

• manual regulation by stage number;

• manual adjustment of the transverse conductivity value;

• automatic voltage regulation;

• planned time for the control device to remain in this operating mode;

• positions of control devices of regulation devices;

• setting value for transverse conductivity of compensating devices;

• voltage setting value of local control devices;

wherein

formation of a data slice and control of data completeness is carried out using a subsystem for optimal control of voltage and reactive power,

obtaining a slice of telemetry and the state of switching devices is carried out through an SQL query to the archive database of the coordinating system of regime automation, archived data slices are formed on the side of the coordinating system of regime automation, and the composition of the source data slices is similar to the composition of the data received operationally, while the source data is supplemented with a label slice time, which is assigned uniformly for all slice parameters,

Next, logical control of the input data is carried out, for which the resulting data slice is checked for compliance with the following conditions:

- if both the power flow and the state of the switching device are measured at the connection, then their compliance is checked;

- if there are measurements of power flows across all connections extending from the buses, then control that the sum of all flows does not exceed the permissible imbalance specified by the setting;

- when sequentially forming slices, data variability is controlled, in accordance with the rule that between two successive slices within one object there must be at least one change in measurements or states;

- presence of signs of unreliability in the source data;

update the initial data for assessing the state, for which they compare the teleinformation with the calculation model made in the calculation module for modeling electrical modes (RMMER) and representing an equivalent circuit of the electric power system, which is equivalent to a real electrical circuit and adequately reflects the processes occurring in it, while in the equivalent circuit, real network elements are replaced by idealized elements, and the equivalent circuit is presented in the form of a graph consisting of nodes and branches connecting them, while a node is a set of connected elements of the same voltage class, having a resistance equal to zero or close to zero, which for this type of calculation can be neglected, and a branch is a section of an electrical circuit that connects two nodes having non-zero resistance, with at least one branch suitable for each node, and the branches are identified by three numbers: “node number of the beginning of the branch”, “node number of the end of the branch” ", "Circuit number", in such a way that if two nodes connect several parallel branches, then the circuit number is set different for each circuit, and for single-circuit lines the circuit number is set equal to zero, while power lines (power lines) in the equivalent circuit are represented by a branch with longitudinal resistance and transverse conductivity,

Then, in the state assessment module, a state assessment is carried out, which consists in obtaining a steady state that would be closest to the existing measurements, while the state assessment module, using static and dynamic methods, performs:

- calculation of all parameters of the current mode and rejection of erroneous teleinformation based on telemeasurement data on the position of switching equipment;

- calculation of retrospective modes based on the archive of telemeasurements in the SSPI;

- checking the reliability of the data of all parameters used in the calculation of the modes;

- preparation of the initial model for other tasks of the complex,

in this case, the static state estimation method is solved independently for each moment in time and produces a balanced steady state for each measurement slice, and the dynamic state estimation method updates its state when moving from one slice to another,

then they perform forecasting of modes based on an artificial neural network (ANN), while processing the results of state assessment by the dynamic method with a modified Kalman filter and accepting them as the final result of state assessment only if the accuracy of the estimates obtained by the dynamic method of state assessment exceeds the accuracy of the estimates obtained a static method of assessing the condition, while the accuracy of the estimates is determined by the value of the total relative deviation,

and in the static state estimation method, the weighted least squares method is used, based on minimizing the following objective function:

- measured mode parameters,

ν (x) - calculated mode parameters based on the values of the system state vector x, and

matrix R _v - covariance matrix, in the absence of correlation between different measurements, representing a diagonal matrix of measurement variances

- the number of measurements, while the result of the state assessment is a state vector containing the estimated values of stresses and stress angles for each node,

- the total number of nodes of the electrical network, while to determine the minimum of the objective function, a system of nonlinear equations is solved, obtained by equating the gradient of the objective function to zero:

to evaluate the circuit, indicate a node that is a reference voltage and with a zero voltage phase, while in the static method of assessing the state, the function of preliminary rejection of erroneous telemeasurements is used, which rejects only voltages in nodes and flows in branches, and loads and generation are not affected by the rejection algorithm,

further, depending on the specified accuracy of telemeasurements, on the magnitude of the unbalance and on the specified threshold of unbalance admissibility, all telemeasurements included in this balance group are recognized as reliable or doubtful, while each telemeasurement has three parameters: measurement sign, measured value and accuracy;

measurement signs are set at the commissioning stage of the isolated power system of the system and adjusted during the operation of the power system, while the measurement signs have the following symbols:

N - “rejected for a long time”, which is used to reject telemetry;

S - “statistical data”, which denotes load and generator nodes for which the load or generation value is set in the basic mode;

# - “rejected”, which is set automatically if the rejection algorithm rejected the measurement, while the measurement accuracy is determined by the total error of the entire measuring path, performed using the same approach,

in this case, the relative deviation of the calculated value of the parameter from its measured value is calculated using the formula

θ _i - measurement variance,

in this case, the relative deviation is a dimensionless relative quantity,

and when analyzing the condition assessment, the total relative deviation is calculated as the sum of all measurements using the following formula:

in this case, nodes in which there is no measurement are not included in the calculation of the total relative deviation; also, rejected measurements, measurements of balancing nodes, measurements in disconnected lines are not included in the calculation of the total deviation; in this case, a combined algorithm is used for calculating the initial value of load power based on load graphs and machine learning using ANN,

and as hierarchical load graphs, a combination of daily graphs and annual coefficients is used.

R _n0 - - initial value of load or generation power in the basic mode;

Р _сг.з - the value of the coefficient of the winter daily active power schedule in relative units for a given point in time;

R _sg.l - the value of the coefficient of the summer daily schedule of active power in relative units for a given point in time;

_Kz - “winter coefficient”, set for each month; in this case, a value equal to 1 corresponds to working only on a winter schedule, 0 - only on a summer schedule,

and the value for a specific day of each month is calculated by a linear approximation over the values for the corresponding months,

the value of the reactive power of the load is calculated using the ratio between active and reactive power tg(φ) specified for each load;

As a result of forecasting using hierarchical load graphs, predicted measurement values and estimates of measurement variances are obtained:

i is the current moment in time, x is the forecasting depth, and the variance σ(τ) depends exponentially on the forecasting depth:

as a result of forecasting modes using an ANN, the vector is obtained as a forecast predicted measurements, and

also the standard deviation of each predicted value from the actual measurement on the test sample:

n _t - number of slices in the test sample,

- test value,

- measured value;

the final indicator of forecast accuracy is the total standard deviation for all predicted σ ² measurements, which is used to calculate the variance and which in turn is used to obtain a generalized forecast result using the state assessment method;

The operation of the forecasting algorithm is monitored by the following information and diagnostic messages:

- forecast mode in the form of diagrams and tables;

- values of relative measurement deviations in the form of diagrams and tables;

- values of measurement dispersions in the form of diagrams and tables;

- generalized indicator of the quality of the updated forecast;

- in case of long-term deterioration in the quality of the updated forecast, an emergency message is issued;

Next, they calculate the limits of static stability, which is an indicator of the possibility of synchronous operation of synchronous electric generators and synchronous motors of the power system with slow load changes, set restrictions for intelligent load control, and also calculate stability margins taking into account the static characteristics of the load in voltage P _n (U, f)P(G), _Qn (U,f) and frequency, while

static load characteristics are specified separately for P and for Q in the form of polynomial coefficients:

- for active load power, coefficients A ₀ , A ₁ , A ₂ , A _f are set to calculate the dependence of load power on voltage and frequency using the formula:

- for reactive load power, set the coefficients V ₀ , B ₁ , V ₂ , B _f to calculate the dependence of load power on voltage according to the formula:

then set operating restrictions, which are input data for the algorithm for optimal control of voltage and reactive power, minimum and maximum restrictions on the voltage level in the nodes of the electrical network, such as:

- restrictions on the injection of reactive power in electrical network nodes, which are generators and compensating devices;

- current limits in the branches of the “nodes/branches” model, corresponding to current limits on power lines and transformers;

- restrictions on the flow of power in the branches of the “nodes/branches” model, corresponding to the restrictions on the power of power lines and transformers,

in this case, operating restrictions are automatically calculated in the intelligent load control cycle by an algorithm for analyzing reserves for static stability and operating limitations by a weighting algorithm and correspond to the operating modes of the power system, the optimization of which for each moment of time is carried out in a deterministic formulation:

And

- elements of the voltage vector U at the nodes of the electrical network, and optimization is performed taking into account the following restrictions on changes in mode parameters:

and form equations for the steady state of the electrical network in the form of equality constraints to the optimization problem:

in this case, if the stability margins are violated, the task of entering the mode into the permissible region of stability with a given safety factor is performed:

δ - violation of stability boundaries,

U ₁ - voltage at load nodes,

the result is formed in the form of a voltage limit for all nodes of the power system:

load characteristics are determined by the following methods:

- active experiment, which is a forced change in voltage in a power system node;

- passive experiment, which is possible only during periods of stationary load;

At the same time, they use intelligent automatic load shedding (ALS) and minimize load shedding due to:

- calculation of the required volume of load shedding directly when issuing a control action;

- dynamic caller ID - issuing impacts in several stages, separated in time, taking into account the dynamics of the accident;

- “flickering” caller ID - a technology for optimizing the shutdown of the heating load, in which part of the load in the power system is turned off for a certain time without any consequences for the consumer, while first of all the control actions of the shock wave are issued to the load connected to the “flickering” caller ID, then after the specified time has elapsed, the load is turned on, and another load of similar volume is turned off, while the total load of the area where the caller ID is implemented is reduced by the required amount, and the time of power supply interruption for each consumer does not exceed the specified value;

and the principle of operation of ANI is that the total volume of power of the switched off load is equal to the volume of power of the control action:

P _YB - volume of control action power,

P _OH - power of one load to be switched off,

i - serial number of the load;

and make the following assumptions:

- disconnected loads brought under the action of AON are grouped into load disconnection stages of fixed power, and the power volume of the stage with a large serial number includes the power volume of the stage with the previous serial number:

P _CTj - the power volume of the load cut-off stage is equal to the power volume of the control action ;

P _j - power step of the switched-off load of the j-th stage;

in this case, the power of each control action is set depending on the configuration of the corresponding load shedding (LO) stages formed in the AON, in turn, the power of the LO stage is equal to the total power of the grouped LO commands, LO subgroups, and the power of the stage with the previous serial number, and the power of the grouped OH commands, OH subgroups, is equal to the total load power of consumers assigned to this command at the substation level in the place where the outage is actually implemented, the load power of the consumer when calculating the power of the control action (CA) is selected based on previously accumulated statistical data, while the operational adjustment function is available consumer load, which is used when the season changes;

and the HC is calculated in two stages:

• at the first stage, in offline mode, when setting up the emergency control automation, a large-scale calculation of the required HCs is performed on the complete circuit of the emergency control power system, which is guaranteed to return the power system to a stable and permissible flow regime;

• at the second stage, a refined calculation of the HC is carried out to disaggregate the impact on load shedding and reduce the total volume of the load to be disconnected; the calculation is carried out for each load node separately;

• calculation of the HC is performed according to the following algorithm:

• before starting the calculation of the shock wave, possible shock waves are specified;

• the magnitude of the load shedding power is determined by the load shedding stage in the node, which is less than or equal to the load value of the consumer connected to the caller ID;

• simulate possible emergency events and for each emergency event simulate predetermined emergency response scenarios;

• if the calculation of the post-emergency mode is successful and the current limits are not violated, then it is considered that it can exist;

• among all successful HCs, the minimum is chosen;

• minimize calculations by sorting HCs from smallest to largest, selecting successful HCs and stopping the calculation;

• a successful shock wave is checked for stability of the dynamic transition;

• as a result of the calculation, we obtain the total volume of outages with a sufficient stability margin;

• at the second stage, on a dynamic model of the power system online in the load shedding executive system (ISON), a refined calculation of the required shock waves is performed, which takes into account the temporary characteristics of the load and is produced and for which the initial circuit for calculating the shock wave is a diagram on which only one load node is presented , while the power system is represented by an equivalent generator, and each load is represented by its real measured value;

the circuit is performed as an equivalent circuit in which the loads are connected directly to one node, while the calculation of each dynamic transition is optimized in order to determine the maximum time for disconnecting a unit load and minimizing the disconnection of loads;

In this case, calculations are performed according to the following algorithm:

• for each shock wave obtained as a result of the calculation at the first stage, the dynamic transition is calculated, and if the impact does not violate the dynamic stability of the power system, then the shock wave is optimized as follows:

• perform a division of the shock, expressed in the disconnection of a certain volume of load, into stages in accordance with the possibility of disaggregating the load by intelligent OH devices;

• perform repeated modeling of the dynamic transition in such a way that the load is not switched off immediately, but in parts with pauses between the issuance of impacts;

• obtain a load division into steps with a given shutdown speed;

HC calculation is performed using the area method, while the initial load power P ₀ and emergency power P _A are known in advance from the first stage of the calculation, performed offline, and the load powers ΔP _i and the minimum shutdown times of each load, the average value of the switched off load are known is defined as:

ΔPi is the magnitude of each load,

n is the number of loads participating in this generalized impact;

in this case, the loads are evenly divided into stages so that the load power of the node changes according to a linear law and the acceleration area is equal to:

Δδ is the change in the generation angle corresponding to the time during which the load is disconnected;

Moreover, in terms of power change, this is equivalent to one effect through time Δt on the value ΔР, and the braking area is equal to:

and the maximum impact angle is equal to:

and using the resulting angle deviation and the constant inertia of the power system, the time during which the angle deviation Δδ occurs is found;

further, real impacts are included in the resulting linear characteristic so that the area does not exceed the linear averaging, while taking into account the margin for the disconnected load and the formation of several mutually redundant load sets, and during trouble-free operation of the power system, the power system is configured and optimized for transmitting commands, for which Pseudo-commands for shutdown are transmitted, the time of command transmission is measured, an adjustment is made for the execution of the command directly on the device, and thus for each load the time characteristics of the load response to the command are formed, then at the upper level the required stages of caller ID shutdown obtained as a result of optimization are correlated with real load shedding stages indicating the maximum shutdown time for each stage,

and the generated stage numbers are transmitted to intelligent load shedding devices, and when a load shedding command arrives, the device disconnects the required load as quickly as possible without additional calculations, while

The data transfer network between Caller ID devices is carried out using technology developed on the basis of p2p technology, which unites Caller ID end devices, ensures the issuance of electricity from electricity consumers into a single Caller ID network without organizing a strictly hierarchical data transmission network, while

information delivery routes are optimized using a multi-agent approach, which works independently of the main data transmission algorithm, moves randomly through the network in the form of a virtual entity and finds a route with the best time characteristics, then this route is marked with the corresponding measured time characteristics, and at the time of implementation of the HF transmission data is carried out along all possible routes, ensuring redundancy and maximum data transfer speed,

and when performing the shock, the actual load shedding is monitored, and control is carried out in the following ways:

• control of the passage and execution of the command,

• control of the measured load value at the lower level, at the level of the disconnecting device,

• control of the load value for the entire consumer;

Moreover, if, as a result of control, the required volume of load shedding is not provided within a given time, then I carry out the further process according to two scenarios:

- disconnecting the entire consumer,

- disabling a fast guaranteed power reserve, a load that is connected to an intelligent Caller ID and has the shortest shutdown time,

in this case, the intelligent load control subsystem (ILC) prepares shock waves, and to obtain optimal shock waves at a given depth, a dynamic optimization algorithm is used, for which the initial data are:

- current evaluated mode,

- forecast states of the power system for the considered time range,

- list of possible HCs,

- state of the hydrocarbons;

the forecast states of the system for the considered time range are the set:

X _n - vector of the system state at each moment of time;

in this case, the cost of control actions depends not only on the state vector of the system, but also on time, and the entire time range is divided into several subranges, each of which contains the optimal shock wave for a given subrange; the complete objective function of dynamic optimization is described by the formula:

f _i (X _i ) - vector of optimal control parameters on the subrange,

b _p and e _p - indices of the beginning and end of the subrange in the time slice,

C is a function of the cost of impacts, depending on the cost of each specific hydrocarbon, taking into account the time specified by the index,

ξ ₀ is taken equal to the initial value of the shock wave,

f _i is the objective function of each dynamic optimization subtask, including:

- active power losses,

- deviations from permissible voltage ranges,

- violation of regime restrictions;

the search for a global optimum over the entire forecast range is carried out taking into account the cost of shock waves, taken into account in the full objective function, and to obtain the optimal value of continuous shock waves, a hybrid method of simulated annealing for continuous quantities and the gradient descent method are used, while the PIUN issues tasks to the software and hardware complex of intelligent control voltage and reactive power (PTK IUNRM), as well as intelligent caller ID settings, containing:

- restrictions in the form of a boundary on the maximum and minimum voltage for nodes, a calculation model used in the software and hardware complex of IUNRM, with each limitation containing several voltage measurement points corresponding to the points of the network at which the voltage is controlled by a local regulator,

- for each restriction, its duration is specified,

- restrictions are hierarchical: if the validity period of a restriction ends, then a restriction of a higher level comes into effect, while it is controlled that the restrictions are not contradictory, both in values and in time, but at the highest level of the hierarchy of boundaries, in the root element hierarchy, basic boundaries with infinite action time are found;

PIUN transmits a set of settings to ISON, which are processed in the event of an emergency using signals from the existing emergency automatic system, ISON transmits to LUUN, which is the connection controller and controls the switching device of a specific, final load, a shutdown command and turns off the load switching device, while

To control the operation of the PIUN from the operator’s side, the following information is displayed on-line at the workstation:

1. Forecast results to a certain depth:

- estimated forecast error,

- the greatest differences between the predicted regime and the current regime,

- predicted regime in the form of diagrams and tables;

2. Statistics on equipment life, maximum and average continuous currents;

3. Results of determining stability reserves;

4. Optimization results:

- optimal mode in the form of diagrams and tables,

- differences between the target function in the initial and initial modes,

- HCs corresponding to this optimal regime;

Moreover, all Caller ID devices operate on the uniform principle of digital Caller ID, which has the following functions:

- formation of dynamic stages of OH,

- formation of a set of load shedding power volumes,

- coordination of control actions to disconnect the load,

- guaranteed shutdown of power volume,

- ensuring a given level of consumption,

- control of the available volume of shutdown power,

- preparation of balancing hydrocarbons for the device for selecting switchable generators,

- ungrouping of the consumer load into priority and secondary,

and use a two-stage caller ID algorithm, in which, if the load fails to be disconnected, the consumer is disconnected using local load control devices (LCDs) using IMS means:

- when a control action (CA) occurs for ANI, the load value for quick shutdown is determined <50 ms, while the remainder is allocated to a separate group for the implementation of LUN shutdown;

- tasks are submitted that include the required OH values depending on the values before emergency consumption for each of the LUUN;

- LUUNs form the necessary list of consumers to be disconnected based on pre-prepared data on consumers;

- a command is sent via communication channels from the LUN to the consumer’s input panel;

- the internal intelligent input device disables the required group;

- LUUN expects a load reduction for a specified time, taking into account both the reduction in load to the expected value and the speed of load reduction;

- if the load reduction is not detected, the entire LUN consumer is disconnected;

- if a decrease is recorded, it means that the ANI IMS tools have worked successfully.

In development of the invention, in the absence of fixing a time stamp on the final measuring device, the time stamp is assigned by an information collection and transmission system (ICS).

In further development of the invention, in the absence of fixing the time stamp on the final measuring device, the time stamp is assigned on the local control device.

In further development of the invention, the teleinformation slices include additional information entered by the dispatcher through the coordinating system of regime automation.

In further development of the invention, the node/branch model is prepared by a topology processor, namely:

- nodes connected by a branch of zero resistance are equivalent;

- the total parameters of nodes, such as load, generation, shunt, are calculated based on the parameters of the power system equipment.

The set of input data for the node/branch model is quite limited. In reality, the system contains equipment with its own passport data. To automatically generate a simplified model from the full model, the task of automatic generation of a computational model (topological processor) is used.

Moreover, it should be noted that simplification of the model is carried out only from the point of view of information presentation. The accuracy of the calculations does not deteriorate at all. In addition, the simplified node/branch model does not contain branches with resistance close to zero. This improves the convergence of the steady state calculation process.

An intelligent load control device (ILOD) in isolated power systems in emergency modes, designed to be integrated into the control system of electric power facilities and containing a set of hardware, an automated workstation (AWS), a set of software, characterized in that the set of hardware consists of a subsystem intelligent load control (PIUN), executive load shedding system (ISON) and local load control devices (LUND), while the PIUN is designed to interact with upper-level control devices - software and hardware complex for intelligent voltage and reactive power control (PTK IUNRM ) to minimize technological losses in the electrical network; and the lower level - by the executive load shedding system (ISON), while the ISON contains intelligent automatic load shedding (ALO) and is designed with the ability to receive control signals from emergency automatics, as well as determine the list of loads to be switched off and transmit a telecontrol signal to disconnect loads to local devices disconnection of loads, and LUUN are designed with the ability to monitor the condition, magnitude of loads, power quality parameters of the electrical installation of a consumer of a separate connection, as well as the ability to disconnect or enable the load.

In further development of the invention, the intelligent load control subsystem (ILC) contains a module for monitoring the condition of the main power equipment, a module for calculating static stability limits, and a module for calculating automatic load shedding settings (ALS).

Recently, the issue of uninterrupted energy supply for both household and industrial consumers has become particularly relevant. For household consumers, uninterrupted energy supply is necessary in the massive transition to heating with electric heat sources. In addition, utilities are also extremely sensitive to power outages. The constant complication of industrial processes also makes production extremely sensitive to power outages. Accordingly, the shutdown time must be minimized or, if possible, shutdown avoided altogether. The claimed invention finds the required balance between the need to prevent an accident using system automation and the undesirable consequences of disconnecting consumers.

The claimed invention contains the principle of dividing the load on the consumer side into several groups depending on the “undesirability” of a shutdown. The principles of grouping can be completely different and depend primarily on the end consumer. So, for houses equipped with alternative heat sources, the most undesirable thing would be to turn off the lights, and vice versa.

A large supply of computing power, significant communication capabilities, as well as a wide territorial coverage of the hardware platform make it possible to create, based on the claimed invention, intelligent control centers for Smartgrid functionality with a mechanism for predicting the occurrence of emergency situations and preventing the development of accidents.

The use of the claimed invention in isolated energy systems by consumers in isolated, hard-to-reach and island territories is especially relevant.

The claimed invention provides prediction of the occurrence of emergency situations and prevention of the development of accidents, which increases the service life of isolated power systems and increases the reliability of power supply to consumers in isolated, hard-to-reach and island territories.

Brief description of drawings

The claimed invention is illustrated in graphic materials. In fig. Figure 1 shows a diagram of an isolated power system with a built-in intelligent load control device.

1 - operational information complex of the power system (OIC);

2 - teleinformation: telemeasurements (TI), telesignals (TS);

3 - software and hardware complex for intelligent control of voltage and reactive power (PTK IUNRM);

4 - automated workstation (AWS);

5 - forecast modes;

6 - current network mode;

7 - transfer of restrictions;

8 - statistics on the service life of power equipment;

9 - intelligent load control subsystem (IPUN);

10 - module for calculating static stability limits;

11 - module for calculating the settings of the intelligent Caller ID;

12 - module for monitoring the condition of the main power equipment;

13 - load profiles;

14 - load shedding volume;

15 - maximum and average currents;

16 - reactive power sources (RPS), reactive power compensators (RPC);

17 - transformers;

18 - generators;

19 - load shedding executive system (ISON)

20 - control actions (CA);

21 - emergency automatics (EA);

22 - state and value of load parameters;

23 - telecontrol (TC), control actions (CF)

24 - local load control device (LUUN);

25 - consumer of a separate connection voltage 0.4, 6-10 kV;

26 - consumer of a separate connection voltage 35 kV;

27 - consumer of a separate connection voltage 110-500 kV;

28 - intelligent load control device (UIUN).

In fig. Figure 2 shows a block diagram of the operation of the subsystem for optimal control of voltage and reactive power of the power system.

201 - collection of information about the current state of the system;

202 - formation of a data slice in the nodes/branches model;

203 - condition assessment;

204 - mode prediction;

205 - calculation of limits on statistical stability and formation of restrictions for subsidiaries;

206 - dynamic optimization (DO) taking into account restrictions;

207 - control of optimization results;

208 - calculation of settings for intelligent Caller ID;

209 - issuance of control actions (CA)

In fig. 4 - Passive experiment, where:

(a) - power supply conditions change greatly, the load changes slightly;

(b) - power supply conditions change slightly, the load changes greatly.

Carrying out the invention

Regulation of the load connected to the electrical network is based on the integration of all energy sources and consumers into a single information space. In this case, each device that is part of the control network is equipped with control components, the operation of which is regulated by software.

The more complex the power system, the more control elements are required, which leads to the need for increased capital investment and labor costs in the design and implementation of such a technical solution. At the moment, there are no devices for integrated load control on the Russian market. To implement an integrated load management system, it is necessary to divide all consumers into groups according to importance, and manage each of them separately, based on priorities. The load management system must automatically calculate energy consumption, and if the value established for a particular consumer is exceeded, return the load to acceptable limits by turning off non-priority groups of consumers or applying a load control effect.

In addition to load priorities, characteristics such as permissible frequencies, minimum and maximum load shutdown times, and forced shutdown duration must also be taken into account.

The use of the claimed invention in isolated power systems in emergency modes increases the service life of isolated power systems, eliminates unnecessary shutdowns of consumer power receivers and minimizes the consequences of emergency disturbances in isolated power systems.

The functioning of the intelligent load control device (ILC) 28 (Fig. 1) in the control system of electric power facilities is provided in conjunction with devices for preventing the occurrence of emergency situations, as well as with blocking control actions (CA) when the emergency automatics (EA) 21 are triggered (Fig. 1), in particular load shedding.

Switching off the load of electrical energy consumers (EE) is used to prevent instability, limit the reduction in frequency and voltage, eliminate overload of controlled sections, power lines (PTL) and equipment. It is carried out by disconnecting all electrical connections of the power receiving installations of electrical energy consumers with the power system with the prohibition of automatic re-enablement and the automatic entry of a reserve of disconnected connections. Power receiving installations of electrical energy consumers of all categories of power supply reliability can be connected to the OH.

When the ES is in operation, the minimum required level of electrical energy consumption in accordance with the level of emergency or technological armor is ensured by the consumer of electrical energy using autonomous backup power supplies with automatic start, provided for by the power supply reliability category of this consumer.

There are also alternatives to disconnecting the load when the power transmission is overloaded:

- Shutdown (or unloading) of generators in the transmission system;

- Emergency increase in generated power in the receiving system;

- Temporary reduction in power consumption due to the regulating effect of voltage load;

- Control of excitation of generators in the receiving system;

- System division (Option: allocation of a certain number of power plant generators to an overloaded transmission).

UIUN 28 (Fig. 1) is intended to regulate the load connected to the electrical network and, in case of exceeding the value established for a particular consumer, return the load to acceptable limits by disconnecting non-priority groups of consumers or applying the regulating effect of the load.

Control of the UIUN 28 (Fig. 1) is performed using an automated workstation (AWS) 4 (Fig. 1) and communication ports for communication with a higher-level automated system using the IEC 61850 communication protocol.

The functionality of UIUN 28 (Fig. 1) includes:

- intelligent load control subsystem (PIUN) 9 (Fig. 1), which works in conjunction with the subsystem for optimal control of voltage and reactive power of the power system (POU IUNRM) (Fig. 2);

- load shedding executive system (ISON) 19 (Fig. 1), which, based on the calculated volume of load shedding 14 (Fig. 1), determines the list of loads to be switched off and, depending on the stage of automatic load shedding (AL), transmits a telecontrol signal 23 (Fig. 1 ) to disconnect loads to local load disconnect devices;

- local load control device (LUUN) 24 (Fig. 1), which controls the states and values of load parameters 22 (Fig. 1) based on teleinformation: power quality parameters of the electrical installation of the consumer of a separate connection 25, 26, 27 (Fig. 1) and upon command from ISON 19 (Fig. 1), it turns off or turns on the load.

This division of UIUN 28 (Fig. 1) into functional subsystems ensures reliable control in real time. Failure of the computationally intensive mathematical algorithms of the intelligent load control subsystem (PIUN) 9 (Fig. 1) does not lead to the failure of the entire isolated power system, but transfers the isolated power system to work according to previously calculated AON settings that provide acceptable control.

PIUN 9 (Fig. 1) ensures interaction with control devices of the upper software and hardware complex for intelligent control of voltage and reactive power to minimize technological losses in the electrical network; PTC IUNRM 3 (Fig. 1) and the lower level ISON 19 (Fig. 1).

PIUN 9 (Fig. 1) calculates stability margins and calculates the settings of the intelligent Caller ID.

PIUN 9 (Fig. 1) consists of three functional modules:

- module for calculating static stability limits 10 (Fig. 1);

- module for calculating the settings of the intelligent Caller ID 11 (Fig. 1);

- monitoring module for main power equipment 12 (Fig. 1).

The PIUN 9 subsystem (Fig. 1) is designed to provide the following functions:

- Calculation of static stability reserves taking into account the regulating effect of the load;

- Calculation and clarification of the values of static load characteristics (SLC);

- Transfer of restrictions 7 (Fig. 1) to the dynamic optimization block of the software and hardware complex IUNRM 3 (Fig. 1);

- Obtaining the results of dynamic optimization from the software and hardware complex IUNRM 3 (Fig. 1);

- Calculation of settings for intelligent Caller ID;

- Transfer of intelligent caller ID settings to ISON 19 (Fig. 1).

ISON 19 (Fig. 1) is functionally an intelligent Caller ID, a decentralized adaptive control system.

The advantage of ISON 19 (Fig. 1) is its action, which is ahead of the work of all other types of automation. In ISON 19 (Fig. 1), taking into account the information received from all LUUN 24 (Fig. 1) and from the upper level of PIUN 9 (Fig. 1), an algorithm is implemented for selecting Caller ID queues for the stages of load shedding and the action to turn on the load. ISON 19 (Fig. 1) generates individual commands to turn off and turn on the load of a specific Caller ID queue in LUUN 24 (Fig. 1). At the output of ISON 19 (Fig. 1), as many commands are generated as there are queues for turning off the load and turning it back on. These commands must be transmitted from ISON 19 (Fig. 1) to LUUN 24 (Fig. 1) via redundant emergency control channels or via the IEC 61850 protocol.

The main task of ISON 19 (Fig. 1) is to shift the functions of Caller ID into the distribution network, disaggregate the OH stages, decentralize the control of PA 21 devices (Fig. 1) and organize control of the previous mode on site.

LUUN 24 (Fig. 1) represents distributed automation that forms the optimal composition of outages and is intended to control the parameters of the consumer's electricity and control its switching in the operating cycle of PA 21 (Fig. 1) (AON).

Each queue of actuators LUUN 24 (Fig. 1) is automatically configured to respond to a specific command to turn off or turn on the load from ISON 19 (Fig. 1).

The functions of LUUN 24 (Fig. 1) in terms of monitoring load operation are as follows:

- control of technological limitations of electrical energy quality parameters;

- control of the total power consumed by the load;

- logging (logging) of all incoming commands from ISON 19 (Fig. 1);

- displaying diagnostic messages about load operation to the user.

The local load control device LUUN 24 (Fig. 1) is a measuring transducer with built-in software.

LUUN 24 (Fig. 1) measures the parameters of the modes of three-phase alternating current electrical networks with a nominal frequency of 50 Hz, and also performs the functions of telecontrol, telesignaling and technical electricity metering, ensuring the exchange of information via galvanically isolated digital interfaces RS-485 and Ethernet.

LUUN 24 (Fig. 1) are intended for use as part of UIUN 28 (Fig. 1) and allow the creation of distributed telemechanics systems and PA 21 (Fig. 1), technical electricity metering systems, monitoring systems for the quality of electrical energy and main energy equipment.

LUUN 24 (Fig. 1) support the transfer of information both directly to PIUN 9 (Fig. 1), and as part of an intelligent caller ID system, telemechanics through a telemechanics server or operational information complex of the power system (OIC) 1 (Fig. 1).

LUUN 24 (Fig. 1) provides measurement and transmission of electrical network mode parameters via RMS communication interfaces:

- root mean square values of alternating current and voltage, active, reactive and apparent powers, active and reactive energy in forward and reverse directions;

- parameters of the electrical network mode based on currents and voltages of the fundamental frequency:

• effective values of alternating current, voltage, active, reactive and apparent powers;

• network frequencies;

• power factors cos ϕ (tg ϕ; ϕ) (phase-by-phase and average);

• individual parameters of power quality - zero-sequence voltage (U0), positive-sequence voltage (U1), negative-sequence voltage (U2), negative-sequence voltage asymmetry coefficient (K2U), sinusoidal distortion coefficient of the voltage curve (KU), zero-sequence current ( I0), positive sequence current (I1), negative sequence current (12), current unbalance factor (K2I), current sinusoidal distortion factor (KI), harmonic distortion factor (THD).

LUUN 24 (Fig. 1) provides control actions (telecontrol) through built-in discrete outputs based on commands received via digital interfaces from ISON 19 (Fig. 1). When the states of any discrete signal change, the events are logged, a timestamp is assigned, and the captured state is ready for transmission. The accuracy of assigning a time stamp is no worse than 1 ms.

LUUN 24 (Fig. 1) is equipped with a real time clock. The presence of a clock allows you to assign uniform astronomical time stamps to entries in the event log and transfer parameters using standard protocols with time stamps. Clock synchronization is carried out using the SNTP protocol.

LUUN 24 (Fig. 1) measures the effective values of currents and voltages, as well as calculates the total, active and reactive power for individual phases, taking into account higher harmonics. To measure the parameters of the electrical network mode (effective values of alternating current and voltage, active, reactive and apparent power, active and reactive energy in the forward and reverse directions), the following well-known expressions are used.

To achieve the required accuracy and operating modes of the device, the following characteristics of analog signal reception circuits are required:

- Measurements are carried out from current and voltage measuring transformers directly without the use of intermediate converters with: In = 1 A; 5 A Unom=57.7 V; 100 V, 220, 380, 690 V;

- Long-term overload capacity of voltage inputs, no less than 1.5 Un (1.8 for the “open delta” input);

- Long-term overload capacity of current inputs, not less, long-term 2 In, short-term (1 s) 40 In;

- Dynamic range of current channels, no less than Inom 0.1-10 (at 50 A at the 500 mV input on the analog-to-digital converter (ADC));

- Dynamic range of the current channel 3I0, no less, Inom 0.1-10 (at 50 A at the input 3.2 V to the MK);

- Consumption per phase in alternating current circuits at Inom and alternating voltage at Unom: no more than 0.5 VA;

- Duration of the measurement and survey cycle, no more than 0.1 s;

- The plausibility of the input analog signals must be checked;

- The limit of the main reduced error should be no more than: ±0.2% or ±0.5%.

The differential mode of operation of the analog-to-digital converter (ADC) of the microcontroller is used, while the matching amplifier has 2 outputs - direct and inverse with the same positive signal offset. This doubles the ADC bit capacity and removes the DC component.

To protect against overvoltages, a varistor with an operating voltage of 470 V is installed. The resistance of the voltage divider resistors is selected taking into account the requirement to ensure the dynamic range of the input. In order to reduce the main reduced error, resistors are used that have a temperature coefficient of resistance of no more than 50ppm/°C. As in current circuits, the differential operating mode of the microcontroller ADC is used

To implement the required operating modes, LUUN 24 (Fig. 1) contains digital inputs.

The digital communication circuits of LUUN 24 (Fig. 1) are equipped with interfaces that allow the integration of devices into substation control systems, the use of convenient methods for setting parameters, and also control via a local computer. To implement the required operating modes of the device, the following characteristics and operating protocols of digital communication circuits are required:

- Number of Ethernet interfaces - 2 pcs.;

- Number of RS-485 interfaces - 1 pc.; Digital communication protocols: Modbus RTU, Modbus TCP, GOST R IEC 870-5-101, GOST R IEC 870-5-103, GOST R IEC 870-5-104; IEC 61850-8.1, SNTP.

The claimed device uses built-in microcontroller and external peripherals and memory connected via serial (UART, SPI, I2C) and parallel buses (FMC).

PIUN 9 (Fig. 1) is designed to maintain the optimal normal mode of the power system according to the stability criterion in real time based on telemetering data and cyclic solution of optimization problems, as well as calculating the settings of Caller ID 11 (Fig. 1).

The intelligent load control subsystem uses the following components of the software and hardware complex IUNRM 3 (Fig. 1):

1. Integration platform providing:

- Receiving teleinformation 22 (Fig. 1) (TI, TS) about the current state of the power system according to the standard protocol IEC 60870-5-104;

- Execution of optimal control algorithms;

- Transfer of control actions to a centralized coordinating system of automatic control systems, ensuring optimal control in accordance with a given target function.

2. Archive database for storing: archives of information about the state of the power system, operating data of the subsystem for optimal control of voltage and reactive power.

3. Message subsystem, which provides messages to the user and logging of diagnostic information about the operation of the system.

Functional modules PIUN 9 (Fig. 1), operating in the control cycle, carry out:

- Calculation of limits on static stability of current and forecast modes.

- Calculation of settings for intelligent Caller ID 11 (Fig. 1).

The calculation module for modeling electrical modes (RMMER), state assessment and optimization solves the following problems:

- Calculation of steady state;

- Optimization of losses;

- Entering the permissible voltage range;

- Condition assessment;

- Calculation of losses;

- Stability analysis;

- Calculation of limiting modes using the weighting method (static stability, maximum permissible modes for currents, voltages, flows);

- Dynamic stability analysis;

- Preparation and work with mnemonic diagrams (structural diagrams, electrical connection diagrams).

Within the framework of PIUN 9 (Fig. 1) of the power system, RMMER performs basic computing functions.

Interaction with RMMER computing modules is carried out using a COM server.

In fig. Figure 2 shows a block diagram of the operation of the optimal voltage control and reactive subsystem:

1. Information about the current state of the power system 201 (Fig. 1) is collected, for which they use an information collection and transmission system (ICTS), which collects information about the state of the power system from the existing operational information complex of the power system (OIC) by initiating the process of preparing measurements suitable for modeling modes and optimization, such as:

- obtaining archived sections of television information and network status, with the help of which the learning process and the initial launch of forecasting are carried out;

- obtaining current sections of television information and network status

- logical control of the integrity of the received data;

- formation of a slice in the “nodes/branches” model 202 (Fig. 2) using a topological processor, while linking telemetry to a single-line diagram of the electrical network.

2. In the SSPI, a teleinformation slice 22 is formed (Fig. 1), which is a set of measurements and telesignals about the state of the network taken over a certain time. Time simultaneity is determined by the time stamps present in each analog measurement or switching device state measurement. The timestamp is typically transmitted from the final digital measuring device (digital converter). If the time stamp is not fixed on the final measuring device, the time stamp is assigned on the collection device (SSPI) or on the local control device. SSPI is used as a collection system, which serves for:

- Collection of current teleinformation according to the IEC 60870-5-104 protocol;

- Formation and storage of sections of television information in the database at a specified frequency;

- Issuing sets of control actions; Maintaining archives of received and issued information.

Composition of TV information sections 22 (Fig. 1):

- Telemetry of active and reactive (P, Q) power along power lines, including lines going to the consumer (supplying the load);

- Voltage measurements on station and substation buses;

- Conditions of switching devices; Operating mode of local control devices:

- Manual regulation by stage number;

- Manual regulation of the transverse conductivity value;

- Automatic voltage regulation;

- Planned time for the control device to remain in this operating mode;

- Positions of the control devices of the control devices (number of the current tap tap, number of the current control stage of the compensating devices during step regulation);

- The setting value of the transverse conductivity of the compensating devices (with continuous regulation);

- The value of the voltage setting of local control devices.

The teleinformation slices 22 (Fig. 1) also include additional information entered by the dispatcher through the coordinating system of operational automation, such as the operating mode of local automation devices.

Formation of a data slice and control of data completeness is carried out using the subsystem for optimal control of voltage and reactive power included in the software PTK IUNRM 3 (Fig. 1)).

Obtaining a snapshot of telemetry measurements and the state of switching devices is carried out through an SQL query to the archive database of the coordinating system of regime automation. Archived data slices are formed on the side of the coordinating system of regime automation.

The composition of the source data slices is similar to the composition of the data received online, except that the data is supplemented with a slice timestamp. The slice timestamp is assigned uniformly for all slice parameters and may differ from the timestamps received from devices.

So, for example, if a slice is formed over time t, but some data has not changed (for example, the states of switching devices) during time θ, then the timestamp received from the end device for this data will be t-θ. The slice timestamp t must be recorded in the slice for this data. In this case, the device timestamp must also be present in the slice as a separate field.

Next, logical control of the input data is carried out. The resulting data slice is checked for compliance with the following conditions:

- if both the power flow and the state of the switching device are measured at the connection, then their compliance is checked;

- if there are measurements of power flows across all connections extending from the buses, then the sum of all flows should not exceed the permissible imbalance specified by the setting;

- with sequential formation of slices, data variability is controlled. In this case, between two consecutive slices within one object there must be at least one change in dimensions or states;

- presence of signs of unreliability in the source data.

In order to update the initial data for assessing state 203 (Fig. 2), teleinformation 22 (Fig. 1): telemetering (TI)/telesignal (TS) is compared with the calculated model RMMER. This comparison is ensured by “linking” TI/TS to the calculation model.

The binding is carried out from the graphical diagram presented in RMMER, and is stored accordingly in the graphical diagram. This approach makes it possible not to increase the size of the calculation model to switching circuits.

Thus, as a result of “binding”, an element of a graphical diagram is associated with one vehicle and several (according to the number of types) TI.

For example, TC state of the switch, flow P through the switch, flow Q through the switch. Linking is carried out using a unique vehicle/TI identifier. The transformation of measurements and telesignals to the model level is carried out by a topological processor.

By a power system model, nodes/branches mean an equivalent circuit of an electrical power system or network that is equivalent to a given electrical circuit and adequately reflects the processes occurring in it. In an equivalent circuit, real network elements (physical devices) are replaced by idealized elements (active resistances, capacitances, inductances, ideal transformers, master currents and powers). The equivalent circuit is represented as a graph consisting of nodes and branches connecting them.

A node is understood as a set of connected elements of the same voltage class, having a resistance equal to zero or close to zero, which can be neglected for this type of calculation. An example of a node can be: substation buses, the location where the tap is connected to the power line.

A branch is a section of an electrical circuit that connects two nodes and, as a rule, has a non-zero resistance. Each node can have from 1 to several branches. Examples of branches: transformers, overhead and cable lines, longitudinal reactors.

Branches are identified by three numbers: “node number of the beginning of the branch”, “node number of the end of the branch”, “node of the chain”. If two nodes connect several parallel branches, then the chain number is set differently for each chain. For single-circuit lines, the circuit number is set to zero. Power transmission lines (PTL) in the equivalent circuit are represented by a branch with longitudinal resistance and transverse conductivity.

The set of input data for the node/branch model is quite limited. In reality, the system contains equipment with its own passport data. To automatically generate a simplified model from the full model, the task of automatic generation of a computational model (topological processor) is used.

Moreover, it should be noted that simplification of the model is carried out only from the point of view of information presentation. The accuracy of the calculations does not deteriorate at all. In addition, the simplified node/branch model does not contain branches with resistance close to zero. This improves the convergence of the steady state calculation process. The node/branch model is prepared by the topology processor, namely:

- Nodes connected by a branch of zero resistance are equivalent;

- Based on equipment data, the total parameters of nodes (load, generation, shunt) are calculated.

3. Software PTK IUNRM 3 (Fig. 1) (software block or, in other words, state assessment module 203 (Fig. 2) - a component of the optimal control subsystem, produces a state assessment, which consists in obtaining such a steady state, which was would be closest to the available measurements.The result of the state estimation task is:

- Calculation of all parameters of the current mode (active and reactive load and generation powers, modules and phases of nodal voltages, flows of active and reactive powers) and rejection of erroneous teleinformation based on telemetering data (TI) and telesignals (TS) about the position of switching equipment;

- Calculation of retrospective modes based on the telemetry archive in the SSPI;

- Reliability of all parameters used in the calculation of modes;

- Preparation of the initial model for other tasks of the complex (steady state, optimization, etc.).

The solution to the state estimation problem 203 (Fig. 2) is carried out by two methods: static and dynamic.

The problem of static state estimation is solved independently for each time instant and allows us to obtain a balanced steady state for each measurement slice.

The task of dynamic state estimation updates its state when moving from one slice to another, thereby taking into account the change in the operating mode of the EPS over time. These two state assessment methods complement each other, verifying the assessment results.

The PTK IUNRM 3 software (Fig. 1) predicts modes based on an artificial neural network (ANN). The results of dynamic state assessment are accepted as the final result of state assessment only if the accuracy of the estimates obtained by dynamic state assessment exceeds the accuracy of estimates obtained by the method of static state assessment. The accuracy of the estimates is determined by the value of the total relative deviation. If the results of dynamic state assessment are better than the static results, then short-term mode forecasting with a depth of up to 1 minute is carried out using dynamic state assessment methods.

The results of short-term forecasting of modes are used to estimate the state at the next time steps. This makes it possible to perform state assessments in the event of a loss or delay in obtaining part of the measurements. Predictive measurements are used with significantly greater variance than measured ones. The amount of change in the variance of the predicted measurements is adjusted during system commissioning. To solve the problem of static state estimation, the weighted least squares method is used. Today, this approach is the classical formulation of the state estimation problem. In the classical formulation, the most common method for solving the state estimation problem is the weighted least squares method, based on minimizing the following objective function:

- measured mode parameters,

v (x) - calculated mode parameters based on the values of the system state vector x.

Matrix R _v - covariance matrix, in the absence of correlation between different measurements, representing a diagonal matrix of measurement variances

- number of measurements.

The result of state estimation 203 (Fig. 2) is a state vector containing the estimated values of stresses and stress angles for each node,

- the total number of nodes in the electrical network.

It should be noted that due to the obvious insufficiency of measurements, so-called pseudo-measurements are also used, that is, expert values of load and generation obtained, for example, based on control measurements or other data.

To determine the minimum of the objective function, a system of nonlinear equations is solved, obtained by equating the gradient of the objective function to zero:

The solution to this system of equations is performed by linearizing the nonlinear equations at each point and using the Newton-Raphson method with choosing the optimal step at each iteration.

To evaluate the circuit, you need to specify a basic node - a node that is the reference voltage and has a zero voltage phase. Since the voltage in the reference node is fixed, it is important to set the reference node to a reliable voltage measurement.

In the static state estimation block, there is a function for preliminary rejection of erroneous measurements: it rejects only voltages in nodes and flows in branches, and loads and generation are not affected by the rejection algorithm.

The state assessment block (a separate software module as part of the software package IUNRM 3 (Fig. 1) automatically compiles groups of TIs in which, with greater or less accuracy, it is possible to check the correct balance of TIs based on Kirchhoff's laws or interconnection of TIs based on the laws of loop stresses. For example, it is possible checking the coincidence of the algebraic sum of measurements of power flows in all branches approaching a node and the measured load of the node; checking the coincidence of measurements of power flows at the beginning and end of a branch with an approximate account of losses and capacitive generation of the branch; checking the coincidence of measurements of flows in parallel branches, etc. This approach is known as the control equation method.

Further, depending on the specified accuracy of the TI, the magnitude of the unbalance and the specified threshold of tolerance of the unbalance, all TI included in a given balance group can be recognized as reliable or doubtful (and intermediate gradations between reliability and doubtfulness can also be introduced).

Each such measurement is specified by three parameters: measurement sign, measured value and accuracy. Signs of measurements are set at the stage of commissioning of the system. In the future, they can be adjusted during its operation. Measurement signs have the following symbols: N - rejected for a long time. This sign can be selected to reject telemetry. S - statistical data. For load and generator nodes for which the load or generation value is set in the basic mode, in the absence of telemetry, the value from the basic mode is substituted. If there is data on load graphs, then statistical data will be substituted taking into account the load graphs. # - rejected. This flag is set automatically if the algorithm for rejecting bad measurements rejects this measurement. Thus, the imbalance in the surrounding part of the network will be “written off” to this node.

The measurement dispersion is specified as the measurement accuracy. It should be taken into account that the accuracy of measurements obtained from TM devices is determined by the total error of the entire measuring path. For greater clarity, standard deviations can be specified as accuracy. This will only affect the value of the objective function. In this case, it is important that for all measurements one approach is used to specify the measurement accuracy (either dispersion or standard deviation). The minimum acceptable accuracy value is 0.1. For most loads, the recommended accuracy value is 20-30% of the original load value, for loads with significant variability (for example, traction railway) - 40-80%.

Recommended accuracy: 1) The accuracy of measuring voltages in nodes is set in accordance with the expertly accepted average accuracy of a voltage transformer of 1%, that is: in a 110 kV network - 1; in the 220 kV network - 2; in a 500 kV network - 5 in nodes with voltage less than 35 kV - 0.5.

2) The accuracy of measuring flows P, Q is set equal to: voltage less than 110 kV - from 0.5 to 1; in the 110 kV network - 2; in the 220 kV network - 4; in a 500 kV network - from 10 to 28.

3) The accuracy of the measured loads/generations is set on average to 2-3% of the maximum load/generation value.

4) For statistically specified (not measured) loads, the accuracy is set from 20 to 30% of their current value for most loads, and for a load that is fully or partially traction - from 40 to 60%.

The relative deviation of the calculated value of a parameter from its measured value is calculated using the formula

θ _i - measurement variance.

Relative deviation is a dimensionless relative quantity.

For loads and generation nodes θ _i with a “+” sign means the presence of excess power in the node in comparison with the measured or statistically calculated value, with a “-” sign - insufficient power.

The practical usefulness of the relative deviation is that it characterizes the degree and severity of the violation of the measured value, regardless of the absolute value of this violation and the specified accuracy, i.e. the value of the relative deviation is an invariant indicator with respect to the specified values. For a group of connected nodes, for individual sections of the network diagram, it usually shows a general tendency for a group of nodes towards excess or lack of power, caused by measurements of power flows in the branches along the boundaries of this group of nodes and their total load and generation. For greater information and clarity, the magnitude of the tendency towards excess or lack of power is also calculated for nodes in which there is no load or generation and is also displayed in the form of a relative deviation.

An important quantity when analyzing condition assessment is the total relative deviation, calculated as the sum of all measurements using the following formula:

Nodes that do not have a measurement do not participate in the calculation of the total deviation (although the relative deviation is calculated for them). Also, rejected measurements, measurements of balancing nodes, and measurements in disconnected lines are not included in the calculation of the total deviation.

Algorithm for calculating the initial value of load power. There are several possible ways to set the initial value of node loads

- There is a TI of active and reactive load power:

- There is a TI of only active power.

The reactive power value will be calculated using the given load tg(φ):

- There is a TI of only the load current. The values of active and reactive power are calculated based on the specified rated load voltage and tg(φ);

- TI loads are missing or found to be erroneous. The initial value of active power is calculated by interpolating between the winter and summer maximum load using the winter-summer coefficient.

This coefficient is set by the corresponding schedule and is automatically adjusted during the calculation process. If the load has a link to one of the standard daily schedules, the above calculations will be performed taking into account such a schedule.

Reactive power will be calculated using tg(φ).

The problem of dynamic state estimation is solved using the Kalman filter. To predict weakly variable components of the system state vector for a short period of time, dynamic state estimation based on modification of the Kalman filter is used. The lead time can be from several seconds (the interval between receiving measurement data (slice) at time k and at time k+1) to 1 minute. The Kalman filter is a classic method for dynamic state estimation.

The dynamic state estimation block does not require entering additional initial data beyond those specified for static state estimation. However, in order for dynamic state assessment to begin to produce adequate results, time is required to accumulate statistics.

If you have an archival database, it is possible to accumulate statistics using retrospective information from it. To monitor the operation of the state estimation algorithm, the issuance of user messages is provided. For generalized control over the operation of the state estimation block, the values of the total relative error of the static and dynamic state estimation algorithms are used.

4. Prediction of modes (Fig. 3) in the subsystem for optimal control of mode and reactive power is performed by a combined algorithm using forecasting based on load graphs and machine learning using artificial neural networks.

Forecasting depth up to 1 day.

A combination of daily schedules and annual coefficients is used as hierarchical load graphs.

R _n0 - initial value of load or generation power in the basic mode.

P _сг.з - the value of the coefficient of the winter daily active power schedule in relative units for a given point in time,

P _sg.l - the value of the coefficient of the summer daily schedule of active power in relative units for a given point in time, K _z - the “winter coefficient” set for each month. A value equal to 1 corresponds to working only on a winter schedule, 0 - only on a summer schedule.

The value for a specific day of each month is calculated by a linear approximation to the values for the corresponding months. The reactive power value of the load is calculated using the ratio between active and reactive power tg(φ) specified for each load. As a result of forecasting using hierarchical load graphs, predicted measurement values and estimates of measurement variances are obtained:

i is the current moment in time, τ is the forecasting depth.

The variance σ(τ) depends exponentially on the prediction depth

It is possible to use refined forecasting using ANN. A recurrent neural network LSTM is used as the basic architecture of the ANN for predicting modes. Using recurrent LSTM networks for load forecasting allows one to obtain results that are not inferior to FCRBM networks.

At the same time, based on LSTM networks, it is possible to build deep networks, which in turn makes it possible to identify more complex dependencies in the source data. This LSTM cell architecture solves the problem of vanishing gradients that has prevented recurrent networks from learning long-term dependencies.

One of the important features of the initial data for solving the problem of optimal control of voltage and reactive power is their probabilistic nature. Input measurements are not simply given by a value, but are also accompanied by variances. This provides additional information for training the neural network. In addition, if at the output of the neural network we have not only the values of the mode parameters, but also the probability distributions for these parameters, then such information will allow us to make more informed decisions on mode control based on the forecasting results. Therefore, Bayesian ANNs are used to solve the forecasting problem. You can build a Bayesian neural network that takes into account the probabilities of the ANN parameters using the Bayes By Backprop method (BBB).

The Bayes by Backprop (BBB) method allows you to obtain the posterior distribution of neural network weights

d is the dimension of the parameter space of the neural network.

This distribution is typically a Gaussian distribution.

The input values of the forecasting algorithm for each considered moment in time are the measured voltage values in the nodes of the electrical network, the flow of active and reactive power in the branches and the measurements of injections of active and reactive power in the nodes (load and generation). Also, if there are vector measurements, the source data may contain measured voltage angles.

The vector of input measurements can be sorted according to the order of nodes that is optimal from a convergence point of view, or it can be unordered.

The archival database of initial data for predicting the mode is filled with a software and hardware complex for intelligent control of voltage and reactive power to minimize technological losses in the electrical network; POU IUNRM. In addition to historical measurement data converted into a node/branch model, the prediction algorithm requires information about measurement variances. This information is prepared during the commissioning process of the condition assessment unit.

The forecasting block using an ANN produces a vector as a forecast predicted measurements, as well as the standard deviation of each predicted value from the actual measurement on the test sample:

n _t - number of slices in the test sample,

- test value,

- measured value.

The final indicator of forecast accuracy is the total standard deviation for all predicted measurements. The standard deviation σ ² is used to calculate the variance, which in turn is used to obtain a generalized forecasting result using the state estimation method. The proposed architecture and forecasting algorithm allows one to reliably tune out the forecasting errors of the refined method. An expert approach using multi-level load graphs gives a more reliable forecast, but with large dispersion. A refined algorithm using an ANN produces a forecast with less variance. In this case, possible fluctuations in the updated forecast are immediately leveled out by calculating the standard deviation (and, accordingly, variance) and subsequent assessment of the state.

To monitor the operation of the forecasting algorithm, the following information and diagnostic messages are provided:

- Forecast mode (on the diagram and in tables).

- Values of relative measurement deviations (in the diagram and in tables).

- Values of measurement dispersion (in the diagram and in tables).

- Generalized indicator of the quality of the updated forecast (total standard deviation

- If there is a long-term deterioration in the quality of the updated forecast, an alarm message is issued.

5. The PTK IUNRM 3 software (Fig. 1) calculates the static stability limits. Static stability means the possibility of synchronous operation of synchronous electric generators and synchronous EPS motors with slow load changes. The task of calculating static stability is assessed on the basis of cyclic calculations of steady-state conditions (SS). The criterion for violation of static stability is the lack of convergence of the UR after a gradual increase in severity of the regime. To find the limiting mode using the weighting method, a weighting trajectory is specified, i.e. a list of nodes in which generation and load changes and the initial change step for each node. Within the framework of the task of intelligent load control, the main task is the formation of a task in the software and hardware complex IUNRM 3 (Fig. 1) in the form of restrictions. The weighting trajectory can be obtained automatically based on the difference between the optimal and the original mode. Moving further along this trajectory using the weighting method, we can obtain the limiting regime, both in terms of static stability and in terms of operational or technological limitations. Calculation of stability margins is carried out taking into account the static characteristics of the load in terms of voltage P _n (U, f) P (G), Q _n (U, f) and frequency.

Static load characteristics are specified separately for P and for Q in the form of polynomial coefficients.

For active load power, coefficients A ₀ , A ₁ , A ₂ , A _f are set to calculate the dependence of load power on voltage and frequency using the formula:

For reactive load power, the coefficients B ₀ , B ₁ , B ₂ , B _f are set to calculate the dependence of load power on voltage using the formula:

The static characteristics of the load in terms of voltage and frequency do not reduce exactly to quadratic polynomials. The voltage dependences are much more complex, and their mathematical description, when presented in polynomial form, must contain a large number of coefficients. The form of presentation of the SCN has to be simplified. Mainly due to the fact that only a small number of coefficients included in the formulas describing SCN can be determined quantitatively. The coefficients of the higher degrees do not have as stable a basis as the coefficients up to the second degree inclusive. However, when describing SCN for the purposes of optimal control, quadratic polynomials are quite sufficient. In the absence of reliable information about the SCN of a power system node, generalized typical coefficients are used, which differ depending on the voltage class and the nature of the simulated load.

6. Mode restrictions are input data for the algorithm for optimal control of voltage and reactive power. The regime restrictions are:

- Limitations on the voltage level in the electrical network nodes (minimum and maximum);

- Restrictions on the injection of reactive power in electrical network nodes, which are generators and compensating devices;

- Current restrictions in the branches of the nodes/branches model, corresponding to the current restrictions on overhead lines and transformers;

- Restrictions on power flow in the branches of the nodes/branches model, corresponding to the restrictions on the power of overhead lines and transformers. Mode restrictions are automatically calculated in the intelligent load control cycle by an algorithm for analyzing reserves based on static stability and mode restrictions by a weighting algorithm. An important feature of operating restrictions is their compliance with the operating mode of the EPS (normal, repair, emergency, post-emergency, forced, etc.).

For unambiguous terminology, the operating mode of the EPS is called the “type” of the mode. For each moment in time, the following optimization problem can be formulated in a deterministic formulation:

u - elements of the voltage vector U at the nodes of the electrical network. Optimization is carried out taking into account the following restrictions on changes in mode parameters:

The steady-state equations of the electrical network are formed in the form of equality constraints to the optimization problem:

In the event of a violation of the stability margins, the task of entering the mode into the permissible region of stability with a given safety factor is performed. To do this, the following optimization problem is solved:

δ - violation of stability boundaries,

U ₁ - voltage at load nodes.

As a result, voltage restrictions are formed for all nodes of the power system:

Methods for experimental determination of load characteristics are divided into two classes: the active experiment method and the passive experiment method (Fig. 3).

An active experiment involves a forced change in voltage in a power system node.

Voltage changes in a fairly narrow range (up to ±10%) are practically not dangerous for consumers and, if technically feasible, can be allowed for any load node.

The adjustment of the SCN is carried out in the process of a passive experiment, which consists in using the variations in frequency and voltage that are always available in the power system. Having registered these variations and the corresponding reaction of the active and reactive load to them, as a result of statistical processing, it is possible to determine the regression lines (JIP) P _n on U and Q _n on U, characterizing their averaged dependencies. In a passive experiment, there should be sufficiently large voltage variations at the node of the measured load due to other loads.

A passive experiment to determine SCN is possible only during periods of stationary load (for example, daily maximum or minimum). At this time, load fluctuations in different nodes are not correlated. If changes in power supply conditions and changes in the parameters of power receivers in the measured load are not correlated, then with the combined action of both of these factors, the measurement data in the P _H (U) coordinates form a certain spot. If these changes obey the normal distribution law, then the spot forms an ellipse.

In the first of these cases, the determination of the LR gives a result close to the line AB, corresponding to the SCN of the measured load (Fig. 3 (a)), but in the second case, the result of the LR does not correspond to the SCN in any way (Fig. 3 (b)).

Intelligent Caller ID (ISON) will minimize load shedding due to:

- calculation of the required volume of load shedding directly when issuing a control action

- dynamic caller ID - issuing impacts in several stages, separated in time, taking into account the dynamics of the accident;

- “flickering Caller ID” - technology for optimizing heating load shutdown.

The principle of operation of AON is that the total volume of power of the switched off load is equal to the volume of power of the control action.

P _YB - volume of control power; P _ON - power of one load to be switched off; i is the number of loads to be switched off (serial number of the load).

To solve this equation in the existing caller ID, the following assumptions are made:

- The disconnected loads brought under the action of AON are grouped into load disconnection stages of fixed power, and the power volume of the stage with a large serial number includes the power volume of the stage with the previous serial number.

P _CTj - power volume of the power cut-off stage; P _j - power step of the switched-off load of the j-th stage. The power volume of the load shedding stage is equal to the power volume of the control action:

The power of each control action is set depending on the configuration of the corresponding OH stages formed in the ANI. The power of the OH stage is equal to the total power of the grouped OH commands (OH subgroups) and the power of the stage with the previous serial number. The power of OH commands (OH subgroups) is equal to the total load power of consumers assigned to a given command at the substation level at the location where the shutdown is actually implemented. When calculating the hydrocarbon power, the load power of the consumer is selected based on previously accumulated statistical data and can be promptly adjusted. Operational adjustment (adjustment of hydrocarbon power volumes) is rarely used, usually when the season changes, due to its high labor intensity. In reality, in practice, due to the low information capacity of the emergency command transmission channels, only pre-grouped substations are available in the central load shedding device, and not individual loads.

This grouping of substages is carried out at places where load shedding commands are received, mostly in places where individual consumers are disconnected.

Collection of information about the input or output of an individual consumer at the level of a substation load shedding device performed by an operational pad or key, as well as the current operational state, mode and readiness of the shutdown circuit is not automated, there is no feedback and requires constant monitoring by operating personnel, and for the central shutdown device and emergency automatics dosing control actions, there may be a need to adjust the settings, requiring constant analysis. From all of the above, it follows that equality (30 is not observed, as a result of which the volume of control actions obviously includes a significant power reserve. This reserve varies depending on the time of day, seasonal period, the nature of the consumer’s technological cycle and in certain situations, taking into account the imposition of failures / faults in the disconnection circuits of an individual consumer, an unacceptable under-disconnection may occur.

To achieve all the benefits of adaptive Caller ID, it is necessary to transfer all Caller ID devices to the uniform principle of digital Caller ID. This will allow you to implement the functions of adaptive caller ID, such as:

- formation of dynamic OH stages or set of load shedding power volume

- coordinated operation of control actions to disconnect the load;

- guaranteed shutdown of power volume or provision of a given level of consumption;

- control of the available volume of shutdown power;

- preparation of balancing hydrocarbons for the device for selecting switchable generators;

- ungrouping the consumer load into priority and secondary.

A large supply of computing power, significant communication capabilities, as well as a wide territorial coverage of the hardware platform make it possible to create intelligent control centers for Smartgrid functionality based on this equipment. It seems possible to solve the following problems:

- Implementation of a “sparing” Caller ID. Those. whenever possible, disconnecting loads that are not critical for the consumer, both in the domestic and industrial consumer segments.

- Smoothing of consumption peaks in EPS. By introducing preferential tariffs for non-priority loads.

- Elimination of overloads in the network of 10 kV and below. With network topology control (reclosers).

- Accounting for reverse generation at the consumer. If the consumer uses alternative energy sources, he can arrange for power to be supplied back to the grid during peak hours. It is also possible in the event of a power shortage in the EPS, for example, at the command of a dispatcher.

Recently, the issue of uninterrupted energy supply for both household and industrial consumers has become particularly relevant. For household consumers, the main reason is the massive transition to heating with electric heat sources, boilers, etc. In addition, utilities are also extremely sensitive to power outages. The constant complication of industrial processes also makes production extremely sensitive to power outages. Accordingly, the shutdown time must be minimized or, if possible, shutdown avoided altogether. The necessary balance between the need to prevent an accident by system automation and the undesirable consequences of disconnecting consumers can be found by implementing selective action of Caller ID.

To do this, it is necessary to divide the load on the consumer side into several groups depending on the “undesirability” of the shutdown. The principles of grouping can be completely different and depend primarily on the end consumer. So, for houses equipped with alternative heat sources, the most undesirable thing would be to turn off the lights, and vice versa.

A variant of a two-stage algorithm for “sparing” Caller ID is proposed. An attempt is made to shutdown using SmartGrid on the consumer side; in case of failure, shutdown using classical means, i.e. the entire feeder at the substation where the LUUN equipment is installed.

The following enlarged Caller ID algorithm is proposed:

- When a control action (CA) occurs for ANI, the load value for quick shutdown is determined <50 ms. The remainder is allocated to a separate group to implement a “gentle” shutdown.

- Tasks are distributed (required OH values depending on the values before emergency consumption) for each of the LUUN devices.

- The devices generate the necessary list of disconnected consumers based on pre-prepared data on consumers.

- A command is sent via communication channels from the LUN to the consumer’s input panel.

- The internal smart input device disables the required group.

- LUUN expects load reduction for some specified time. Moreover, it is possible to both expect a reduction in consumption to the expected value, and take into account the rate of load reduction on the feeder.

- If the decrease is not detected, the entire feeder is switched off.

- If a decrease is recorded, it means the algorithm worked successfully.

When preparing control actions, emergency automation calculates the actions in two stages. At the first stage, an enlarged calculation of the required impact on the complete diagram of the emergency control area is performed. As a result, we obtain anti-emergency effects that are guaranteed to return the system to a stable and permissible flow regime. At the second stage, a refined calculation of the control action is made to disaggregate the impact on load shedding and ultimately reduce the total volume of the load to be turned off. This calculation is carried out for each load node separately.

In practical calculations, the calculation of emergency actions is reduced to the following algorithm. Before starting the calculation of emergency actions, possible control actions are set. The magnitude of the impact on load shedding is determined by the load shedding stage in the node, which, as a rule, is less than or equal to the load value of the feeder installed under the caller ID. Next, possible emergency events are simulated and for each emergency event, predetermined emergency response scenarios are simulated. If the calculation of the post-emergency mode is successful and the current limits are not violated, then it is considered that it can exist. Among all successful emergency measures, the minimum one is selected. You can minimize calculations by sorting the emergency actions from least to greatest and if the emergency actions are successful, then the calculation is not performed further. Successful anti-emergency action is also tested for the stability of the dynamic transition. The calculation of emergency actions at the first stage, as a rule, is carried out offline when setting up the emergency automation. As a result of the calculation, we obtain the total volume of shutdown. This shutdown volume makes it possible to eliminate a possible accident, but has a fairly large margin of stability, which leads to excessive load shutdown. Refined calculation of the required emergency measures.

The next stage is a refined calculation of the required emergency response, which takes into account the temporary characteristics of the load and is carried out on a dynamic model of the power system. These calculations are proposed to be carried out online in the ISON system. The initial diagram for calculating control actions is a diagram that shows only one load node. The system is represented by an equivalent generator, and each load is represented by its actual measured value.

This circuit is equivalent to a circuit in which the loads are connected directly to a single node. The calculation of each dynamic transition is optimized to determine the maximum time for disconnecting a unit load and minimizing the disconnection of loads.

The calculation algorithm is as follows:

- For each emergency response obtained as a result of the calculation at the first stage, a dynamic transition is calculated. If the impact does not violate the dynamic stability of the system, then it is optimized as follows.

- The impact (expressed in disconnecting a certain amount of load) is divided into stages in accordance with the possibility of load disaggregation by intelligent OH devices.

- Next, the dynamic transition is re-modeled in such a way that the load is not switched off immediately, but in parts with pauses between the issuance of impacts. As a result of the calculation, we obtain a division of the load into stages with a given shutdown speed.

This calculation can be done quite simply using the area method. The initial load power P ₀ and emergency power P _A are known in advance from the first stage of the calculation, performed offline. The load power ΔP _i and the minimum shutdown times for each load are known. It should be noted that at this stage a clear breakdown into specific loads is not required. It is important to understand the discreteness of the load. The average switched load is determined as:

ΔPi - the magnitude of each load n - the number of loads participating in this generalized impact. Since the number of unit loads in a node is quite large, and we evenly divided the loads into stages, we can conditionally assume that the load power of the node varies according to a linear law.

In this case, the acceleration area will be equal to

Δδ - change in the generation angle corresponding to the time during which the load is switched off. Moreover, in terms of power change, this will be equivalent to one effect after time Δt on the value ΔР. The braking area is equal to:

How do we determine the maximum impact angle:

Knowing the angle deviation and the constant inertia of the system, you can find the time during which the angle deviation Δδ occurs. After the maximum time for processing the impact has been obtained, we enter the real impacts into the resulting linear characteristic so that the area does not turn out to be larger than the linear averaging. If there is a reserve for the load to be switched off, then to ensure redundancy, several mutually redundant load sets can be formed. During normal operation of the system (when there are no accidents), the system is configured and optimized for transmitting commands. Pseudo-commands for shutdown are transmitted, the time of command transmission is measured, and an adjustment is made for the execution of the command directly on the device. Thus, for each load we can form the time characteristics of its response to the command. After this, at the upper level, the required GA shutdown stages obtained as a result of optimization are correlated with the actual load shutdown stages, indicating the maximum shutdown time for each stage. The generated stage numbers are transmitted to intelligent load shedding devices. Thus, when a command to disconnect the load arrives, the device is ready to disconnect the required load as quickly as possible without additional calculations. To implement a data transmission network between Caller ID devices, a technology developed on the basis of p2p technology is proposed. This technology makes it possible to combine caller ID end devices, which provide control actions to electricity consumers, into a single caller ID network without organizing a strictly hierarchical data transmission network. Advantages of this approach:

- Ability to use any hardware platform for data transmission in different parts of the network.

- Ease of connecting electricity consumers to Caller ID. You just need to connect to any segment of the Caller ID data network and establish a connection with one of the participants in this network (any device on the network).

This could be, for example, a caller ID device at another nearby consumer. Further configuration and optimization of data transmission routes is performed automatically. In pre-emergency mode, information is transmitted in the direction of control (to the head caller ID). Information is combined into packages that can contain data not only from one terminal caller ID device, but also combined packages of information. In this case, the information transmission route can also be optimized. All information packets are accompanied by time stamps, and thus we can monitor the speed of information delivery and optimize the delivery of information based on traffic and speed by enabling or disabling certain routes.

To optimize information delivery routes, a multi-agent approach is used. In this case, the agent is an algorithm (“inspector”) that works independently of the main data transmission algorithm.

By transmitting test messages between network devices, the “inspector” moves randomly through the network in the form of a virtual entity and finds a route with the best timing characteristics. Subsequently, this route is marked with the corresponding measured time characteristics. At the moment the impact is realized, data is transferred along all possible routes, ensuring redundancy and maximum data transfer speed.

When performing control, the actual load shedding is monitored. Control is carried out in several ways: control of the passage and execution of a command, control of the measured load value at the lower level (at the level of the disconnecting device) and control of the load value throughout the entire feeder. If, as a result of control, the required shutdown volume is not achieved within a given time, then the further process can develop according to two scenarios:

- disconnection of the entire feeder;

- disabling fast guaranteed power reserve.

By fast guaranteed power reserve we mean a load that is connected to an intelligent Caller ID and has the shortest shutdown time. Moreover, it may not necessarily be located on this feeder. The second method seems to be the most rational, since under-disconnection of the load may be associated with incorrect load measurements on a given feeder. Thus, even turning it off entirely may not provide the required load.

At certain hours, the energy system experiences a significant increase in consumption - “load peaks”. The presence of consumption peaks leads to the need to maintain a “spinning reserve” at power plants. In addition, recently the load in the household sector, rural areas, etc. has been growing significantly, and the existing network infrastructure does not have time to modernize at such a pace, so overload of equipment (transformers, lines) poses a significant problem. A generally accepted global trend for solving the problem of peaks is to provide consumers with time-varying tariffs that reflect the cost of electricity over different periods of time.

The claimed invention is implemented by installing a multi-tariff electricity meter. An alternative is to install the smart device discussed earlier. The occurrence of a local overload in this case is controlled by the LUN according to a predetermined setting. When an overload is detected, the LUUN issues a command to output certain groups of loads. When the peak passes, the loads are reintroduced. The consumer receives a more preferential tariff by agreeing to install such a system. In addition, it is possible to develop a device to control overloads of the problem area of the EPS, and “gently” disconnect the necessary load. The most promising method of solving the above problems is the installation of energy storage devices and alternative generation sources at the consumer's site. This allows you to remove problems without deteriorating the quality of life, as well as reduce the environmental load. In addition, it seems interesting to organize the reverse supply of power to the network. The created distributed Caller ID network can provide assistance in managing such devices. For example, manage the supply of power to the network. In normal mode, the consumer’s generation can operate “for future use” by accumulating energy in storage devices (batteries) and, if necessary, upon command from the LUUN, release the accumulated power to the network. Due to the fact that a long-term shutdown of even part of the consumer’s load is still undesirable and significantly worsens the consumer’s quality of life, an algorithm for alternately disconnecting various consumers is proposed. If the total value of the disconnected load allows for alternate disconnection of consumers, the LUN determines that the “comfortable” disconnection period has been exceeded and disconnects the next consumer (consumer group) and then switches on the previous one. The “flickering” Caller ID is especially relevant due to the fact that the share of the heating load is increasing.

The principle of flickering Caller ID is quite simple. Part of the load in the power system can be switched off for a certain period of time without any consequences for the consumer. Accordingly, if there is a need to disconnect the load, first of all, impacts can be issued on the load connected to the “flickering” Caller ID. After the specified time has expired, the load will be turned on, while another load of similar volume will be turned off. Thus, the total load of the area where GA is implemented will be reduced by the required amount, but at the same time, the time of interruption of power supply for each consumer will not exceed the specified value.

Implementation algorithm:

- For a load that can be turned off by the “flickering” Caller ID algorithm, the time for which it can be turned off is specified, as well as the minimum amount of electricity that the load should consume (if it is turned on) as a percentage of current consumption.

- The issuance of an influence on the intelligent load shedding device must be accompanied not only by the amount by which the load must be disconnected, but also by the duration of the load shedding.

The duration of the shutdown should take into account the transmission time for shutting down another load in the “flickering” Caller ID cycle.

- Commands to disconnect the load using the “flickering” Caller ID principle will be transmitted in turn to different groups of loads until the cause of the accident is eliminated.

Such an algorithm can work at any level of the Caller ID hierarchy and for almost any type of impact.

PIUN carries out the preparation of control actions (calculation of the required emergency actions). In the intelligent load control subsystem, a dynamic optimization algorithm is used to obtain optimal control actions at a given depth.

The initial data for dynamic optimization are:

- Current rated mode.

- Forecast in the form of a set of predicted regimes.

- List of possible control actions (CA).

- Conditions of the hydrocarbons.

The forecast states of the system for the considered time range are the set:

X _n is the vector of the system state at each moment in time.

The proposed solution takes into account the cost of control actions, which depends not only on the state vector of the system, but also on time. As a result of optimization, the entire time range under consideration is divided into several subranges, on each of which the optimal solution for this subrange is found. The complete objective function of the optimization problem can be written as:

vector of optimal control parameters on the subrange;

b _p and e _p - indices of the beginning and end of the subrange in the time slice;

C is a function of the cost of impacts depending on the cost of each specific impact, taking into account the time specified by the index;

ξ ₀ is taken equal to the initial value of the control parameters.

f _i is the objective function of each optimization subtask, including:

1. active power losses;

2. deviations from permissible voltage ranges;

3. violation of regime restrictions.

This formulation of the dynamic optimization problem allows us to obtain the optimal value for the entire period of time under consideration. The search for a global optimum over the entire forecast range is carried out taking into account the cost of the shock wave, taken into account in the general objective function.

To find the optimal value of continuous control quantities, such as injections of active and reactive power in the nodes of the electrical network graph, it is proposed to use a hybrid method of simulated annealing for continuous quantities and the gradient descent method.

The intelligent load control subsystem issues tasks to the IUNRM software package, as well as intelligent Caller ID settings containing:

- Limitations in the form of a voltage limit (maximum and minimum) for nodes, a calculation model used in the software and hardware of IUNRM.

- Each limitation can contain several voltage measurement points corresponding to network points at which the voltage is controlled by a local regulator.

- For each restriction, its duration is specified.

- Constraints are hierarchical. If a restriction expires, a restriction at a higher level takes effect. At the same time, it must be controlled that the restrictions are not contradictory, both in value and in time. At the highest level of the boundary hierarchy (the root element of the hierarchy) there are basic boundaries with an infinite validity period.

1. The task of changing the voltage level in the nodes for the PTC IUNRM.

2. Calculated values of settings for intelligent Caller ID.

PIUN transmits a set of settings to the ISON (required emergency actions - AON stages), which will be worked out in the event of an emergency based on signals from the existing Emergency Automation system.

ISON transmits the shutdown command to the LUUN, which is the bay controller and controls the switching device of a specific final load. LUUN disconnects the load switching device.

To control the operation of the intelligent load control subsystem on the operator’s side, the following information is displayed online at the workstation:

1. Forecast results to a certain depth:

- Estimated forecast error;

- The greatest differences between the predicted regime and the current regime;

- Predicted regime (in the diagram and in tables).

2. Statistics on equipment life (maximum and average continuous currents).

3. Results of determining stability reserves.

4. Optimization results

- Optimal mode (in the diagram and in the tables);

- Differences between the target function in the initial and initial modes;

- Control actions corresponding to this optimal mode.

Claims (203)

1. A method for intelligent load control in isolated power systems in emergency modes, in which: collect information about the current state of the power system, for which they use an information collection and transmission system (ICTS), which accumulates information about the state of the power system from the existing operational information complex of an isolated power system (OIC) by initiating the process of preparing measurements of modes suitable for modeling and optimization, such as : - obtaining archived sections of teleinformation and network state, with the help of which machine learning and the initial launch of the forecasting process are performed; - receiving and storing in the database sections of television information with a given frequency of current sections of television information and network status; logical control of the integrity of the received data using a set of control actions; - formation of a data slice in the “nodes/branches” model, while linking the teleinformation to a single-line diagram of the electrical network, while the teleinformation slice is a set of telemeasurements and telesignals about the state of the network, taken for a certain time, and the simultaneity of time is determined by the time stamps present in each measurement, while the time stamp is transmitted from the final digital measuring device, while the teleinformation slices include: - telemetry of active and reactive (P, Q) power along power lines, including lines going to the consumer; - voltage measurements on station and substation buses; state of switching devices; - operating mode of local control devices: • manual regulation by stage number; • manual regulation of the transverse conductivity value; • automatic voltage regulation; • planned time for the control device to remain in this operating mode; • positions of control devices of regulation devices; • setting value for transverse conductivity of compensating devices; • voltage setting value of local control devices; in this case, the formation of a data slice and control of data completeness is carried out using a subsystem for optimal control of voltage and reactive power, obtaining a slice of telemetry and the state of switching devices is carried out through an SQL query to the archive database of the coordinating system of regime automation, archived data slices are formed on the side of the coordinating system of regime automation, and the composition of the source data slices is similar to the composition of the data received operationally, while the source data is supplemented with a label slice time, which is assigned uniformly for all slice parameters, Next, logical control of the input data is carried out, for which the resulting data slice is checked for compliance with the following conditions: - if both the power flow and the state of the switching device are measured at the connection, then their compliance is checked; - if there are measurements of power flows across all connections extending from the buses, then control that the sum of all flows does not exceed the permissible imbalance specified by the setting; - when sequentially forming slices, data variability is controlled, in accordance with the rule that between two successive slices within one object there must be at least one change in measurements or states; - presence of signs of unreliability in the source data; update the initial data for assessing the state, for which they compare the teleinformation with the calculation model made in the calculation module for modeling electrical modes (RMMER) and representing an equivalent circuit of the electric power system, which is equivalent to a real electrical circuit and adequately reflects the processes occurring in it, while in an equivalent circuit, real network elements are replaced by idealized elements, and the equivalent circuit is represented as a graph consisting of nodes and branches connecting them, while a node is a set of connected elements of the same voltage class having a resistance equal to zero or close to zero, which for this type of calculation can be neglected, and a branch is a section of an electrical circuit that connects two nodes that have non-zero resistance, with at least one branch suitable for each node, and the branches are identified by three numbers: “Node number of the beginning of the branch”, “Node number end of branch", "circuit number" in such a way that if two nodes connect several parallel branches, then the circuit number is set different for each circuit, and for single-circuit lines the circuit number is set equal to zero, while power lines (power lines) are represented in the equivalent circuit branch with longitudinal resistance and transverse conductivity, Then, in the state assessment module, a state assessment is carried out, which consists in obtaining a steady state that would be closest to the existing measurements, while the state assessment module, using static and dynamic methods, performs: - calculation of all parameters of the current mode and rejection of erroneous teleinformation based on telemeasurement data on the position of switching equipment; - calculation of retrospective modes based on the archive of telemeasurements in the SSPI; - checking the reliability of the data of all parameters used in the calculation of the modes; - preparation of the initial model for other tasks of the complex, in this case, the static state estimation method is solved independently for each moment in time and produces a balanced steady state for each measurement slice, and the dynamic state estimation method updates its state when moving from one slice to another, then they perform forecasting of modes based on an artificial neural network (ANN), while processing the results of state assessment by the dynamic method with a modified Kalman filter and accepting them as the final result of state assessment only if the accuracy of the estimates obtained by the dynamic method of state assessment exceeds the accuracy of the estimates obtained a static method of assessing the condition, while the accuracy of the estimates is determined by the value of the total relative deviation, and in the static state estimation method, the weighted least squares method is used, based on minimizing the following objective function: , Where - measured mode parameters, ν (x) are the calculated mode parameters based on the values of the system state vector x, and the matrix R _v is a covariance matrix, which, in the absence of a correlation between different measurements, is a diagonal matrix of measurement variances - the number of measurements, while the result of the state assessment is a state vector containing the estimated values of stresses and stress angles for each node, - the total number of nodes of the electrical network, while to determine the minimum of the objective function, a system of nonlinear equations is solved, obtained by equating the gradient of the objective function to zero: to evaluate the circuit, indicate a node that is a reference voltage and with a zero voltage phase, while in the static method of assessing the state, the function of preliminary rejection of erroneous telemeasurements is used, which rejects only voltages in nodes and flows in branches, and loads and generation are not affected by the rejection algorithm, further, depending on the specified accuracy of telemeasurements, on the magnitude of the unbalance and on the specified threshold of unbalance admissibility, all telemeasurements included in this balance group are recognized as reliable or doubtful, while each telemeasurement has three parameters: measurement sign, measured value and accuracy; measurement signs are set at the commissioning stage of the isolated power system of the system and adjusted during the operation of the power system, while the measurement signs have the following symbols: N - “rejected for a long time”, which is used to reject telemetry; S - “statistical data”, which denotes load and generator nodes for which the load or generation value is set in the basic mode; # - “rejected”, which is set automatically if the rejection algorithm rejected the measurement, while the measurement accuracy is determined by the total error of the entire measuring path, performed using the same approach, in this case, the relative deviation of the calculated value of the parameter from its measured value is calculated using the formula: where θ _i is the measurement variance, in this case, the relative deviation is a dimensionless relative quantity, and when analyzing the condition assessment, the total relative deviation is calculated as the sum of all measurements using the following formula: , while nodes in which there is no measurement are not included in the calculation of the total relative deviation, also rejected measurements, measurements of balancing nodes, measurements in disconnected lines are not included in the calculation of the total deviation, and a combined algorithm for calculating the initial value of load power based on graphs is used load and machine learning using ANN, and as hierarchical load graphs, a combination of daily graphs and annual coefficients is used: where P _n0 is the initial value of load or generation power in the basic mode; Р _сг.з - the value of the coefficient of the winter daily active power schedule in relative units for a given point in time; P _sg.l - the value of the coefficient of the summer daily active power schedule in relative units for a given point in time; _Kz - “winter coefficient”, set for each month; in this case, a value equal to 1 corresponds to working only on a winter schedule, 0 - only on a summer schedule, and the value for a specific day of each month is calculated by linear approximation based on the values for the corresponding months, the value of the reactive power of the load is calculated using the ratio between active and reactive power tg(φ) specified for each load; As a result of forecasting using hierarchical load graphs, predicted measurement values and estimates of measurement variances are obtained: , Where i is the current moment in time, x is the forecasting depth, and the variance σ(τ) depends exponentially on the forecasting depth: as a result of forecasting modes using an ANN, the vector is obtained as a forecast predicted measurements, as well as the standard deviation of each predicted value from the actual measurement on the test sample: , Where n _t - number of slices in the test sample, - test value, - measured value; the final indicator of forecast accuracy is the total standard deviation σ ² for all predicted measurements, which is used to calculate the variance and which in turn is used to obtain a generalized forecast result using the state assessment method; The operation of the forecasting algorithm is monitored using the following information and diagnostic messages: - forecast mode in the form of diagrams and tables; - values of relative measurement deviations in the form of diagrams and tables; - values of measurement dispersions in the form of diagrams and tables; - generalized indicator of the quality of the updated forecast; - in case of long-term deterioration in the quality of the updated forecast, an emergency message is issued; Next, they calculate the limits of static stability, which is an indicator of the possibility of synchronous operation of synchronous electric generators and synchronous motors of the power system with slow load changes, set restrictions for intelligent load control, and also calculate stability margins taking into account the static characteristics of the load in voltage P _n (U, f)P(G), _Qn (U,f) and frequency, while static load characteristics are specified separately for P and for Q in the form of polynomial coefficients: - for active load power, coefficients A ₀ , A ₁ , A ₂ , A _f are set to calculate the dependence of load power on voltage and frequency using the formula: - for reactive load power, set the coefficients V ₀ , B ₁ , V ₂ , B _f to calculate the dependence of load power on voltage according to the formula: then set operating restrictions, which are input data for the algorithm for optimal control of voltage and reactive power, minimum and maximum restrictions on the voltage level in the nodes of the electrical network, such as: - restrictions on the injection of reactive power in electrical network nodes, which are generators and compensating devices; - current limits in the branches of the “nodes/branches” model, corresponding to current limits on power lines and transformers; - restrictions on the flow of power in the branches of the “nodes/branches” model, corresponding to the restrictions on the power of power lines and transformers, in this case, operating restrictions are automatically calculated in the intelligent load control cycle by an algorithm for analyzing reserves for static stability and operating limitations by a weighting algorithm and correspond to the operating modes of the power system, the optimization of which for each moment of time is carried out in a deterministic formulation: , Where u - elements of the voltage vector U at the nodes of the electrical network, and optimization is performed taking into account the following restrictions on changes in mode parameters: and form equations for the steady state of the electrical network in the form of equality constraints to the optimization problem: in this case, if the stability margins are violated, the task of entering the mode into the permissible region of stability with a given safety factor is performed: , Where δ - violation of stability boundaries, U ₁ - voltage at load nodes, the result is formed in the form of a voltage limit for all nodes of the power system: , load characteristics are determined by the following methods: - active experiment, which is a forced change in voltage in a power system node; - passive experiment, which is possible only during periods of stationary load; At the same time, they use intelligent automatic load shedding (ALS) and minimize load shedding due to: - calculation of the required volume of load shedding directly when issuing a control action; - dynamic caller ID - issuing impacts in several stages, separated in time, taking into account the dynamics of the accident; - “flickering” caller ID - a technology for optimizing the shutdown of the heating load, in which part of the load in the power system is turned off for a certain time without any consequences for the consumer, while first of all the control actions of the shock wave are issued to the load connected to the “flickering” caller ID, then after the specified time has elapsed, the load is turned on, and another load of similar volume is turned off, while the total load of the area where the caller ID is implemented is reduced by the required amount, and the time of power supply interruption for each consumer does not exceed the specified value; and the principle of operation of ANI is that the total volume of power of the switched off load is equal to the volume of power of the control action: Where P _YB - volume of power of the control action, P _OH - power of one switched off load, i - serial number of the load; and make the following assumptions: - disconnected loads brought under the action of AON are grouped into load disconnection stages of fixed power, and the power volume of the stage with a large serial number includes the power volume of the stage with the previous serial number: , Where P _CTj - the power volume of the load cut-off stage is equal to the power volume of the control action P _j - power step of the switched-off load of the j-th stage; in this case, the power of each control action is set depending on the configuration of the corresponding load shedding (LO) stages formed in the AON, in turn, the power of the LO stage is equal to the total power of the grouped LO commands, LO subgroups, and the power of the stage with the previous serial number, and the power of the grouped OH commands, OH subgroups, is equal to the total load power of consumers assigned to this command at the substation level in the place where the outage is actually implemented, the load power of the consumer when calculating the power of the control action (CA) is selected based on previously accumulated statistical data, while the operational adjustment function is available consumer load, which is used when the season changes; and the HC is calculated in two stages: • at the first stage, in offline mode, when setting up the emergency control automation, a large-scale calculation of the required HCs is performed on the complete circuit of the emergency control power system, which is guaranteed to return the power system to a stable and permissible flow regime; • at the second stage, a refined calculation of the HC is carried out to disaggregate the impact on load shedding and reduce the total volume of the load to be disconnected; the calculation is carried out for each load node separately; • calculation of the HC is performed according to the following algorithm: - before starting the calculation of the shock wave, possible shock waves are specified; - the magnitude of the load shedding power is determined by the load shedding stage in the node, which is less than or equal to the load value of the consumer connected to the caller ID; - simulate possible emergency events and for each emergency event simulate predetermined emergency response scenarios; - if the calculation of the post-emergency mode is successful and the current limits are not violated, then it is considered that it can exist; - among all successful HCs, the minimum is chosen; - minimize calculations by sorting the HC from smallest to largest, choosing a successful HC, and stop the calculation; - a successful shock wave is checked for stability of the dynamic transition; - as a result of the calculation, we obtain the total volume of outages with a sufficient margin of stability; • at the second stage, on a dynamic model of the power system online in the load shedding executive system (ISON), a refined calculation of the required shock waves is performed, which takes into account the temporary characteristics of the load and is produced and for which the initial circuit for calculating the shock wave is a diagram on which only one load node is presented , while the power system is represented by an equivalent generator, and each load is represented by its real measured value; the circuit is performed as an equivalent circuit in which the loads are connected directly to one node, while the calculation of each dynamic transition is optimized in order to determine the maximum time for disconnecting a unit load and minimizing the disconnection of loads; In this case, calculations are performed according to the following algorithm: • for each shock wave obtained as a result of the calculation at the first stage, the dynamic transition is calculated, and if the impact does not violate the dynamic stability of the power system, then the shock wave is optimized as follows: • perform a division of the shock, expressed in the disconnection of a certain volume of load, into stages in accordance with the possibility of disaggregating the load by intelligent OH devices; • perform repeated modeling of the dynamic transition in such a way that the load is not switched off immediately, but in parts with pauses between the issuance of impacts; • obtain a load division into steps with a given shutdown speed; HC calculations are performed using the area method, while the initial load power P ₀ and emergency power P _A are known in advance from the first stage of the calculation, performed offline, while the load powers ΔP _i and the minimum shutdown times of each load, the average value of the switched off load are known defined as: , Where ΔPi is the magnitude of each load, n is the number of loads participating in this generalized impact; in this case, the loads are evenly divided into stages so that the load power of the node changes according to a linear law and the acceleration area is equal to: , Where Δδ is the change in the generation angle corresponding to the time during which the load is disconnected; Moreover, in terms of power change, this is equivalent to one effect through time Δt on the value ΔР, and the braking area is equal to: and the maximum impact angle is equal to: , and using the resulting angle deviation and the constant inertia of the power system, the time during which the angle deviation Δδ occurs is found; further, real impacts are included in the resulting linear characteristic so that the area does not exceed the linear averaging, while taking into account the margin for the disconnected load and the formation of several mutually redundant load sets, and during trouble-free operation of the power system, the power system is configured and optimized for transmitting commands, for which Pseudo-commands for shutdown are transmitted, the time of command transmission is measured, an adjustment is made for the execution of the command directly on the device, and thus for each load the time characteristics of the load response to the command are formed, then at the upper level the required stages of caller ID shutdown obtained as a result of optimization are correlated with real load shedding stages indicating the maximum shutdown time for each stage, and the generated stage numbers are transmitted to intelligent load shedding devices, and when a load shedding command arrives, the device disconnects the required load as quickly as possible without additional calculations, while The data transfer network between Caller ID devices is carried out using technology developed on the basis of p2p technology, which unites Caller ID end devices, ensures the issuance of electricity from electricity consumers into a single Caller ID network without organizing a strictly hierarchical data transmission network, while information delivery routes are optimized using a multi-agent approach, which operates independently of the main data transmission algorithm, moves randomly through the network in the form of a virtual entity and finds a route with the best time characteristics, then this route is marked with the corresponding measured time characteristics, and at the time of implementation of the data transfer is carried out along all possible routes, ensuring redundancy and maximum data transfer speed, and when performing the shock, the actual load shedding is monitored, and control is carried out in the following ways: • control of the passage and execution of the command, • control of the measured load value at the lower level, at the level of the disconnecting device, • control of the load value for the entire consumer; Moreover, if, as a result of control, the required volume of load shedding is not provided within a given time, then I carry out the further process according to two scenarios: - disconnecting the entire consumer, - disabling a fast guaranteed power reserve, a load that is connected to an intelligent Caller ID and has the shortest shutdown time, in this case, the intelligent load control subsystem (ILC) prepares shock waves, and to obtain optimal shock waves at a given depth, a dynamic optimization algorithm is used, for which the initial data are: - current evaluated mode, - forecast states of the power system for the considered time range, - list of possible HCs, - state of the hydrocarbons; the forecast states of the system for the considered time range are the set: , Where X _n - vector of the system state at each moment of time; in this case, the cost of control actions depends not only on the state vector of the system, but also on time, and the entire time range is divided into several subranges, each of which contains the optimal shock wave for a given subrange; the complete objective function of dynamic optimization is described by the formula: , Where f _i (X _i ) - vector of optimal control parameters on the subrange, b _p and e _p - indices of the beginning and end of the subrange in the time slice, C is a function of the cost of impacts, depending on the cost of each specific hydrocarbon, taking into account the time specified by the index, ξ ₀ is taken equal to the initial value of the shock wave, f _i is the objective function of each dynamic optimization subtask, including: - active power losses, - deviations from permissible voltage ranges, - violation of regime restrictions; the search for a global optimum over the entire forecast range is carried out taking into account the cost of shock waves, taken into account in the full objective function, and to obtain the optimal value of continuous shock waves, a hybrid method of simulated annealing for continuous quantities and the gradient descent method are used, while the PIUN issues tasks to the software and hardware complex of intelligent control voltage and reactive power (PTK IUNRM), as well as intelligent caller ID settings, containing: - restrictions in the form of a boundary on the maximum and minimum voltage for nodes, a calculation model used in the software and hardware complex of IUNRM, with each limitation containing several voltage measurement points corresponding to the points of the network at which the voltage is controlled by a local regulator, - for each restriction, its duration is specified, - restrictions are hierarchical: if the validity period of a restriction ends, then a restriction of a higher level comes into effect, while it is controlled that the restrictions are not contradictory, both in values and in time, but at the highest level of the hierarchy of boundaries, in the root element hierarchy, basic boundaries with infinite action time are found; PIUN transmits a set of settings to ISON, which are processed in the event of an emergency using signals from the existing emergency automatic system, ISON transmits to LUUN, which is the connection controller and controls the switching device of a specific, final load, a shutdown command and turns off the load switching device, while To control the operation of the PIUN from the operator’s side, the following information is displayed on-line at the workstation: • Forecast results at a certain depth: - estimated forecast error, - the greatest differences between the predicted regime and the current regime, - predicted regime in the form of diagrams and tables; • Statistics on equipment life, maximum and average continuous currents; • Results of determining stability reserves; • Optimization results: - optimal mode in the form of diagrams and tables, - differences between the target function in the initial and initial modes, - HCs corresponding to this optimal regime; Moreover, all Caller ID devices operate on the uniform principle of digital Caller ID, which has the following functions: - formation of dynamic stages of OH, - formation of a set of load shedding power volumes, - coordination of control actions to disconnect the load, - guaranteed shutdown of power volume, - ensuring a given level of consumption, - control of the available volume of shutdown power, - preparation of balancing hydrocarbons for the device for selecting switchable generators, - ungrouping of the consumer load into priority and secondary, and use a two-stage caller ID algorithm, in which, if the load fails to be disconnected, the consumer is disconnected using local load control devices (LCDs) using IMS means: - when a control action (CA) occurs for ANI, the load value for quick shutdown is determined <50 ms, while the remainder is allocated to a separate group for the implementation of LUN shutdown; - tasks are submitted that include the required OH values depending on the values before emergency consumption for each of the LUUN; - LUUNs form the necessary list of consumers to be disconnected based on pre-prepared data on consumers; - a command is sent via communication channels from the LUN to the consumer’s input panel; - the internal intelligent input device disables the required group; - LUUN expects a load reduction for a specified time, taking into account both the reduction in load to the expected value and the speed of load reduction; - if the load reduction is not detected, the entire LUN consumer is disconnected; - if a decrease is recorded, it means that the ANI IMS tools have worked successfully. 2. The method according to claim 1, characterized in that in the absence of fixing a time stamp on the final measuring device, the time stamp is assigned by an information collection and transmission system (ICS). 3. The method according to claim 1, characterized in that if there is no fixation of the time stamp on the final measuring device, the time stamp is assigned on the local control device. 4. The method according to claim 1, characterized in that the teleinformation slices include additional information entered by the dispatcher through the coordinating system of automatic control systems. 5. The method according to claim 1, characterized in that the “nodes/branches” model is prepared by a topological processor, namely: - nodes connected by a branch of zero resistance are equivalent; - the total parameters of nodes, such as load, generation, shunt, are calculated based on the parameters of the power system equipment. 6. An intelligent load control device (ILOD) in isolated power systems in emergency modes, designed to be integrated into the control system of electric power facilities and containing a set of hardware, an automated workstation (AWS), a set of software, characterized in that the set of hardware consists from the intelligent load control subsystem (PIUN), the load shutdown executive system (ISON) and local load control devices (LOUN), while the PIUN is designed to interact with upper-level control devices - the software and hardware complex for intelligent voltage and reactive power control (PTK) IUNRM) to minimize technological losses in the electrical network; and the lower level - by the executive load shedding system (ISON), while the ISON contains intelligent automatic load shedding (ALO) and is designed with the ability to receive control signals from emergency automatics, as well as determine the list of loads to be switched off and transmit a telecontrol signal to disconnect loads to local devices disconnection of loads, and LUUN are designed with the ability to monitor the condition, magnitude of loads, power quality parameters of the electrical installation of a consumer of a separate connection, as well as the ability to disconnect or enable the load. 7. The device according to claim 6, characterized in that the intelligent load control subsystem (ILC) contains a module for monitoring the condition of the main power equipment, a module for calculating static stability limits, and a module for calculating automatic load shutdown settings (ALS).

Images