WO2012143424A1

WO2012143424A1 - System and method for controlling an electrical energy production installation

Info

Publication number: WO2012143424A1
Application number: PCT/EP2012/057130
Authority: WO
Inventors: Mounir DAGGAG; Robin ROCHE; Lhassane IDOUMGHAR; Benjamin Blunier; Abdellatif MIRAOUI; Abderrafiaa KOULAM
Original assignee: Ge Energy Products France Snc
Priority date: 2011-04-19
Filing date: 2012-04-19
Publication date: 2012-10-26
Also published as: FR2974431A1; FR2974431B1

Abstract

System (1) for controlling an electrical energy production installation comprising at least one component, in which at least one control agent (2) of the installation, able to define a total cost of operation of the installation, a required level of electrical production of the installation, and able to issue settings of electrical energy production of at least one component, at least one low-level agent (3, 4, 5, 6), receiving a production or electrical energy consumption setting from the control agent (2) of the installation and issuing a cost value destined for said control agent, each low-level agent (3, 4, 5, 6) being able to control autonomously a component of the electrical energy production installation so as to achieve the electrical energy production setting.

Description

System and method for controlling an electric power generation installation

The invention relates to the technical field of control of industrial installations, and more particularly the control of power generation facilities.

Existing power plants have a monolithic structure based on a single type of energy production (gas, coal, nuclear) and in which we can generally modulate production only to a limited extent. In addition, existing plants are not designed to integrate the cost of polluting discharges into their operation.

Finally, existing plants operate on a tight flow regime in which production must be immediately spent. In situ storage of electrical energy is neither widespread nor integrated into the management of power plants.

Standards for greenhouse gas emissions are becoming more stringent under the pressure of environmental considerations. The creation of new power plants, and the upgrading of existing plants therefore requires taking into account the ecological parameter. For this, smart power plants are developed, in which information technologies play a decisive role.

A first object of the invention is an electrical energy production control system capable of optimizing the production of electrical energy from different components.

Another object of the invention is a control system adapted to take into account pollutant emissions in the management of the production facility.

Another object of the invention is a control system capable of taking into account faults or equipment changes. The invention relates to a control system of an electrical energy production installation comprising at least one component, characterized in that it comprises

at least one control agent of the installation, able to define a total cost of operation of the installation, a required level of electrical production of the installation, and able to issue instructions for the production of electrical energy; at least one component,

at least one low level agent, receiving a production or power consumption instruction from the installation control agent and transmitting a cost value to said control agent,

each low level agent being able to autonomously control a component of the electrical energy production installation in order to reach the electrical energy production target.

A low level agent may be able to communicate the current operating parameters and changes in the characteristics of the controlled component to a plant control agent.

An installation control agent may be able to apply dynamic variation limitation rules to the operating parameters of the components, the plant control agent may also be able to apply a statistical optimization of the production instructions of the plant. electrical energy of the components.

The statistical optimization can be of the PSO type.

The statistical optimization can be modified PSO type by application of the Metropolis rule.

The control system may include an agent register in which each agent declares information about it, maintains this information during its operation, and deletes the information about it during its own deletion. The component for which instructions are issued may be a turbine, a low level agent that may be able to control the starting, stopping and operating level of the turbine according to the characteristics of the turbine, the cost associated with the operation of the turbine depending on the fuel consumption.

The cost associated with the operation of the turbine may depend on polluting emissions.

The component may consist of at least one electrical energy storage means, a low level agent that can be adapted to control the charging and discharging of the storage means depending on the characteristics of the storage means, its charge level. current, the cost associated with a load variation of the storage means depending on the amount of electrical energy consumed.

The component may still be a connection to the distribution network, whereby a low level agent may be able to control the connection of the installation to the distributed distribution network in order to sell the electrical energy produced, the cost associated with a sale of the electricity generated electrical energy depending on the quantity of energy sold and the financial compensation received per unit of energy sold.

The installation may be connected to a set of subscribers, to which electrical power is provided, a low level agent may be able to communicate to the plant control agent the current level of energy demand. by the subscribers, and at least two demand levels at two subsequent predetermined times, the cost associated with supplying electrical energy to the subscribers depending on the amount of energy supplied.

Another aspect of the invention is a method of controlling an electric power generation plant comprising at least one component, comprising steps in which:

- a total cost of operation of the installation and a required level of electrical production of the installation are defined,

the current operating parameters of each component are determined, rules for limiting dynamic variations are applied to the current operating parameters of each component,

a statistical optimization of the instructions for the production of electrical energy of the components is carried out,

the instructions for generating electrical energy from at least one component are issued.

The statistical optimization can be of the PSO type.

Other objects, features and advantages of the invention will become apparent on reading the following description, given solely by way of nonlimiting example and with reference to the appended drawings in which:

- Figure 1 illustrates the main elements of a control system according to the invention, and

- Figure 2 illustrates the main steps of a control method according to the invention.

An electrical generating facility for supplying electric power to an electric power distribution network for supplying it to final customers, for example, comprises one or more gas turbines, one or more energy storage units. electric and at least one connection to the distribution network. Other forms of energy can also be taken into account. The electrical energy supplied to subscribers may pass through a private distribution network or a distribution network shared by several companies.

Figure 1 illustrates a control system 1 of such an installation. The control system 1 comprises several agents 3, 4,5, 6 each able to control a part of the electrical production installation.

An agent is a hardware system, a software method or a combination of such a system and such a method, capable of controlling autonomously the part of the installation to whose control it is assigned.

An agent is able to interact with other agents through the exchange of messages, measurements or instructions. An agent may also interact with his environment, for example, by receiving measurements from sensors, or by issuing operating instructions to systems constituting a part of his environment.

In the example described here, the installation itself and each turbine, each storage means, each connection to the distribution network are each controlled by an agent 2,3,4,5,6.

The control agent 3 of a gas turbine receives the characteristics of the turbine that it controls during its initialization or after maintenance. In real time, periodically or deferredly, the control agent 3 receives values of the current operating parameters of the turbine. These parameters may include in particular the speed of rotation of the turbine, the amount of gas admitted, the amount of electrical energy produced, the temperature of the gases leaving the turbine, the temperature of the air admitted into the turbine. It also receives from the outside, a set of electrical energy to produce and possibly a cost set to be respected.

The control agent 3 modifies the operating parameters of the turbine autonomously, in order to reconcile the electrical energy production instruction by maintaining normal operating parameters, and possibly, by minimizing the operating cost. The cost of operation is related to the amount of fuel consumed. The cost can also take into account the maintenance costs generated by operations in different ranges of values, as well as the costs associated with the production of liquids.

The control agent 4 of a means for storing the electrical energy behaves in a manner similar to that of a turbine. The control agent 4 knows the specifications of the means of storage of the electrical energy, receives values of measuring means characterizing the operating values of the storage means. The control agent 4 receives from outside, a set of electrical energy to destock or store and possibly a cost set to be respected. He then decides on charging or discharging the storage means of the electrical energy. The operating cost may depend on the energy required to increase the state of charge of the storage means, maintenance costs, cycling costs or a combination of these parameters.

The control agent 5 of the distribution network connection controls the opening or closing of the connection to the distribution network, in order to sell electrical power to third parties who are not subscribers.

The subscriber demand control agent 6 determines the amount of electrical energy required by the subscribers, as well as two subsequent two-time estimates, for example 15min and 30min.

The control agent 2 of the installation communicates with the other agents 3,4,5,6, sets their objectives and instructions according to the information they transmit to it. The control agent 2 of the facility maintains a database of active agents and their main characteristics.

Launch Agent 7 creates and deletes a command agent, both at startup and during operation of the control system. The creation and deletion of agents is usually controlled by the control agent 2 of the installation.

The agents can be created as control methods supported by a programming and interfacing language, for example JADE (acronym for Java Agent DEvelopment). The actual creation of agents and their programming depends on the language used and is based on general knowledge of the skilled person. Therefore, these steps are not described here. The production and consumption of electrical energy are distributed among the various components of the installation when executing a control method of the installation.

"Component" means any device or device of the installation capable of performing one or more elementary functions of the installation, such as production of electrical energy, storage of electrical energy, connection to the network, etc.

The control agent 2 of the installation issues the operating instructions, in particular the production of electrical energy, to the various components of the installation. For this, it applies the control method illustrated in FIG.

The method begins with a step 8 during which the control agent 2 of the installation receives low level agents 3,4,5,6, also called control agents of the various components of the installation, the parameters operating the controlled components.

In a step 9, the active low level agents 3,4,5,6 and the control agent 2 declare their existence in the agent register. When declaring an agent, the characteristics of the agent and the equipment it controls are recorded in the agent register. Alternatively, the agent register may be created prior to step 8. Although created in process step 9, the agent register is updated in real time as soon as a change in the stored information occurs. Changing the stored information may be the source of the low level agent or the command agent 2 interacting with the low level agent whose information is being logged.

When the agent whose information is recorded in the register is deleted, for example if the turbine associated with it is declared out of service, in maintenance or simply stopped, the agent removes his information from the register of agents. This has the effect that the control agent 2 is informed of the absence of the deleted agent. Otherwise, the control agent 2 would be confronted with a lack of response from the suppressed agent, whose causes could be multiple and difficult to identify.

The agent register thus contributes to the flexibility of the system by allowing the addition and removal of agents during operation without the need for process adaptation.

The control method is continued by a step 1 0 in which a set of rules differing depending on the nature of the component concerned. The results of step 10 are repeated in a subsequent step 11 to undergo a set of optimizations.

The control rules of a turbine take into account the different modes in which the turbines can operate. These rules, three in number, are as follows:

- In start mode, the maximum and minimum power output levels are set to 0 MW.

- In operating mode, the maximum and minimum power output levels are set to meet the dynamic maxima of the turbine. In other words, the level of electric power production can only increase or decrease by a given value over a given period. The maximum rate of change is thus fixed.

- In case of stoppage, the maximum and minimum electrical energy production levels are adjusted so that the turbine stops as quickly as possible while respecting its dynamic constraints.

The control rules of the electrical energy storage means differ according to whether the subscriber demand is satisfied or not.

If the subscriber demand is satisfied, and the state of charge of the electrical energy storage means is at a load level, the electrical energy storage means is allowed to use electrical energy to to charge.

- If the demand of the subscribers is satisfied, and the state of charge of the electrical energy storage means is at a level not not allowing charging, the electrical energy storage means is not allowed to use electrical energy to charge.

- If the subscriber demand is not satisfied, and the state of charge of the electrical energy storage means is at a level allowing the discharge, the electrical energy storage means is allowed to supply power. electrical energy to discharge.

- If the subscriber demand is not satisfied, and the state of charge of the electrical energy storage means is not at a level allowing the discharge, the electrical energy storage means is not authorized to supply electrical energy to discharge.

Alternatively, in the event that the state of charge of the storage means is too low, it is possible to force the turbines to produce more than their setpoint so that the excess electrical energy allows to reduce the state of charge of the cells. storage means, at a level of equilibrium. The reverse behavior is applied in case of too high state of charge of the electrical energy storage means. This preventive management makes the means of storage available in any situation.

The control rules of the connection to the distribution network differ according to whether the subscriber demand is satisfied or not:

- If the demand of the subscribers is satisfied by the turbines, it is allowed to sell energy via the connection.

- If the demand of the subscribers is not satisfied by the turbines, it is not allowed to sell energy via the connection.

The result of applying the rules described above is then optimized. The rules make it possible to limit the operating zone of each component. Optimization is used to define the operating instructions for each component.

The purpose of the optimization is to minimize the running costs, as defined by the cost function c _to tai: ^C total /, ^C s + Σc _storage - £ c ^ _d (Eq 1)

With

n _t the number of turbines,

n _s the number of storage facilities, and

n _g the number of connections to the distribution network.

The total cost c _to tai is the result of the sum of the cost of each turbine c _tU rbine, cost storage means c _st0 rage, which is deducted the profit from the resale of electricity on the network c _gr id .

The cost of each turbine c _tU rbine depends on fuel consumption, but may also depend on the amount of inches lluants

(eg NOx nitrogen oxides) produced.

The cost of a storage _medium depends on the state of charge of the storage means and the power required. Alternatively, the cost of a storage means can be arbitrarily set equal to zero.

The benefit of the resale of electrical energy is analogous to a negative cost, and corresponds to the product of the quantity of energy sold by the unit cost.

In optimization, cost minimization is constrained by constraints. The constraints include the stability of the network and the respect of the dynamic constraints on the components of the installation.

The network stability constraint ensures that enough electrical power is generated to satisfy subscriber demand. The stability of the network is characterized by equation 2.

y, P nrbine ^ - ^ storage ^- ^ request ~ * ~ T ^{^} , Pprid (Eq.) n, n _s n _g

Pturbine is the power produced by a turbine.

Pstorage corresponds to the power entering or leaving the storage means. The energy is counted positively when a storage means discharges, and negatively when a storage means is charged. Pgrid is the electrical energy sold on the distribution network. This value is always negative, the electric energy can only be sold.

When initializing the optimization, it is estimated that the only known value is the requested power P -

The other constraint corresponds to the respect of the dynamic constraints of the components of the installation.

Thus, for each parameter of each component, a maximum rate of variation is defined over a given period of time. For example, for the power P of a turbine, we write the following equation 3:

P ^r t + l - P ^r t ≤ΔΡ_ (Eq 3)

Has

With

P _t : the power at the instant t

P _t + i: the power at time t + 1

At: a given period of time

AP _max : a maximum variation of power

The actual optimization is based on a Metropolis Particle Swarm Optimization (MPSO) algorithm.

Particle Swarm Optimization (PSO) was developed in 1995 by James Kennedy and Russel Eberhart. It is based on the principle of improving an individual by observing other members of the group and imitating the best.

A particle k is described by a position xk, a velocity vk, and a measure of the performance of the particle.

A standard PSO algorithm follows the following procedure for each particle:

a new speed vk + 1 is determined as a function of the previous speed vk according to equation 4 described below,

a new position xk + 1 is determined according to the previous position xk and the current speed,

- the performance is determined, - the best memorized position is updated.

These instructions are repeated until the maximum number of iterations is reached or the minimum error criterion is reached.

Each particle stores the history of positions in the solution space using two values. The first value is the best solution reached so far by this particle. This first value is called best personal value, noted pbest.

The second value is the best solution reached so far by a particle in the vicinity of this particle. The second value is called the best value of the group, denoted gbest. vk + 1 = wt · vk + cl · rl · (pbest - xk) + c2 · r2 · (gbest - xk) (Eq. 4)

Equation 4 shows that the determination of a new speed depends not only on the previous value of speed, but also on a cognitive component and a social value. For the cognition value, we use the value pbest defined above while for the social component, we use the previous gbest value. The relative importance of the values pbest and gbest depend on the weighting coefficients c l and c2. The relative importance of the velocity vk depends on the weighting coefficient wt. In other words, the displacement of a particle is influenced by the following three components:

• A physical component: the particle tends to follow its current direction of movement;

• A cognitive component: the particle tends to move towards the best site by which it has already passed; and

• A social component: the particle tends to rely on the experience of its peers and, thus, to move towards the best site already reached by its neighbors.

To allow the algorithm to search for solutions, and to ensure that particles are not trapped in minima lo cal, we use two random values rl and r2 generated between 0 and 1.

The MPSO algorithm is a variant of the PSO algorithm described above and has three features. The first is to update the pbest value according to the Metropolis rule. This rule makes it possible to obtain the probability p to accept the current position of the particle as pbest. Equation 5 described below illustrates the rules for calculating this probability p.

1.

0 otherwise

With

f the performance function

x the current particle

r3 a random number between 0 and 1

T temperature

There are several methods for determining the initial temperature among which we can cite the method based on the observation of the average variation of the function f.

From an initial solution xo, a number of solution x 'o (about 100) such that f (x' o)> f (xo) is drawn. So the average variation (delta (f)) is calculated. The initial temperature To is calculated so as to initially accept a certain percentage (p ~ 80% equivalent to a probability p = 0.8) of movement degrading the function f. The value of To is thus deduced from the expression

f-delta (f))

The decay of the temperature used is given by the following geometrical formula: Tk + 1 = y * Tk

The Metropolis rule thus makes it possible to accept "bad" positions during the first iterations, but only the Positions that have contributed to the improvement of the solution are accepted in the last iterations.

The second particularity is how the speed is updated. Unlike the PSO algorithm whose social variable only depends on the gbest value, the social variable of the MPSO algorithm is based on a weighted average of all solutions that are better than the one considered. This defines a new equation 6 for updating the speed based on equation 4 relating to the PSO algorithm.

vk + 1 = wt · vk + cl · rl · (pbest - xk) (Eq.6)

With

vmax the maximum speed allowed,

Nk a set of solutions better than the particle present, ranked in descending order with respect to the objective function, that is to say with respect to the cost function in the context of the application to an electrical production facility.

In other words, equation 6 means that it is better for a particle to follow a group of particles than an isolated particle, the isolated particle can go in a non-optimal direction.

The third particularity concerns the application of a mutation operator if gbest is not improved at the end of a given number of generations, for example 60 generations.

The mutation operator modifies the speed of the particles to their maximum allowed, to allow the algorithm to get out of a situation locked in a local minimum.

The MPSO method is applied by the control agent 2 of the installation during the execution of the control method, each particle corresponding to a component controlling agent.

It is recalled that a particle corresponds to a solution of the problem. Here it is, for a requested power, to find the operating power for each turbine.

At the end of the optimization step 11, we obtain a set of production instructions which are sent to the components in step 12. The process continues, in step 13, by the production instructions being performed by the low level agents controlling the components affected by said setpoints.

The process then continues with a loop to step 9. The process is interrupted only by a suppression of the agents by the launching agent 7.

The electrical installation control system and the associated control method make it possible to optimize the operating costs of an electrical production installation. With production forecasts, the facility control officer 2 can determine whether future demand can be met with the operating means of production. If this is not the case, other means of production are available. If the request is transitory, the energy storage means are solicited in addition to the means of production. In addition, the connection to the distribution network can be cut in order to favor the production of energy to satisfy the demand of the subscribers.

The control system also has sufficient responsiveness to overcome turbine failure by starting other turbines and soliciting the storage means to maintain production during startup of the replacement turbines.

The control system and method described above can be generalized to other sources of electrical energy, not involving a turbine. These energies do not generally involve intermediate conversion by a steam cycle. Examples of these include solar photovoltaic energy, or solar energy.

The control system and method has increased flexibility, responsiveness and efficiency over existing systems.

In addition, these systems and control methods, by their modularity, can be applied to facilities of radically different sizes.

Claims

Control system (1) for an electrical energy production installation comprising at least one component, characterized in that it comprises:

at least one control agent (2) of the installation, able to define a total running cost of the installation, a required level of electrical production of the installation, and able to issue energy production instructions at least one component,

at least one low-level agent (3,4,5,6) receiving a set-point for the production or consumption of electrical energy of the control agent (2) of the installation and emitting a cost value to destination of said control agent,

each low-level agent (3,4,5,6) being able to autonomously control a component of the electrical energy production facility in order to reach the electrical energy production target.

2. Control system according to claim 1, wherein a low level agent (3,4,5,6) is able to communicate the current operating parameters and the changes in the characteristics of the controlled component to a control agent (2). ) of the installation,

a control agent (2) of the installation is able to apply rules for limiting dynamic variations to the operating parameters of the components,

the control agent (2) of the installation being also able to apply a statistical optimization of the electrical energy production instructions of the components.

3. Control system according to claim 2, wherein the statistical optimization is of the PSO type.

4. Control system according to claim 3, wherein the statistical optimization is modified PSO type by application of the Metropolis rule.

5. Control system according to any one of the preceding claims, comprising a register of agents in which each agent declares information about him, maintains this information during its operation, and removes the information about it during its own deletion.

6. Control system according to any one of the preceding claims, wherein at least one component is a turbine, a low level agent (3, 4,5, 6) being able to control the starting, stopping and turbine operating level depending on the characteristics of the turbine, the cost associated with the operation of the turbine depending on the fuel consumption.

7. Control system according to claim 5, wherein the cost associated with the operation of the turbine depends on the polluting emissions.

8. Control system according to any one of the preceding claims, wherein at least one component is at least one electrical energy storage means, a low level agent (3, 4,5, 6) being able to control the charging and discharging of the storage means according to the characteristics of the storage means, its current charge level, the cost associated with a charge variation of the storage means depending on the amount of electrical energy consumed.

9. Control system according to any one of the preceding claims, wherein at least one component is a connection to the distribution network, a low level agent (3, 4, 5, 6) being able to control the connection of the installation in the distribution network used to sell the electrical energy produced, the cost associated with the sale of electrical energy depending on the quantity of energy sold and the financial consideration received per unit of energy sold.

Control system according to one of the preceding claims, in which the installation is connected to a set of subscribers, to which electrical energy is supplied, a low level agent (3, 4,5, 6). ) being able to communicate to the control agent (2) of the installation the current level of demand for electrical energy by the subscribers, and at least two levels of demand at two subsequent predetermined times, the cost associated with the supply of electrical energy to subscribers depending on the amount of energy supplied.

1 1. A method of controlling an electric power generation plant comprising at least one component, characterized in that it comprises steps in which:

the current operating parameters of each component are determined,

rules for limiting dynamic variations are applied to the current operating parameters of each component,

a statistical optimization of the instructions for the production of electrical energy of the components is carried out, and

12. Control method according to claim 11, wherein the statistical optimization is of the PSO type.

13. The control method according to claim 12, wherein the statistical optimization is modified PSO type by application of the Metropolis rule.