RU2316872C1 - Method for relay protection of an energy object - Google Patents

Abstract

FIELD: relay protection of electric systems or any other energy objects.

SUBSTANCE: in accordance to the invention, structure of relay protection consists of two modules of different types; first one generates activation signal, and second one generates protection blocking signal. Protection passes a cycle of training from imitation models of energy object in modes which are alternative to control modes. Protection system features two introduced hierarchical series of additional modules of first and second types. Sensitivity of first ones is increased with advancement toward the end of series, and false action of protection is prevented by training modules of second type to block alternative modes, which cause activation of first type modules. In a specific case, training process occurs without division of already available protection onto two types of modules, but in that case additional modules of second type may operate with new information outside the existing protection, which is replicated by programmed method in this case, and for each following copy more sensitive settings are selected compared to previous one.

EFFECT: expanded functional capabilities.

2 cl, 11 dwg

Description

The invention relates to electrical engineering and electric power industry, namely to relay protection and automation of electric systems or any electric power objects. The first analogue of this invention can be considered the well-known Bresler relay [1], endowed with the ability to process all input quantities, and not just two, as in a conventional resistance relay. The development of the idea of combining all available information was remote ways of protecting power lines [2, 3], which ultimately led to ideas about algorithmic (virtual) relays placed in the branches of the alleged damage in the algorithmic model of a power facility. Virtual relays operate independently of each other. It turns out that the concept of combining information is only half realized by them. The measurement of each virtual relay is determined as the result of the conversion of all available information. Measurements differ in conversion operations. But since the relays operate autonomously and do not secure each other, they have to be so rude that the recognition ability of the protection turns out to be much lower than the recognition of short circuits [5].

A technical solution is known that fundamentally solves the problem of increasing the recognition ability of a power facility protection [6]. It consists in creating a set-point space (measurement space) in the form of a plurality of set-up planes, in dividing each plane into separate cells, in coding sets of cells of different planes, and in training for protection to operate from codes that correspond to controlled modes, and in no case to operate from codes corresponding to alternative controlled modes. It is fundamentally important that in relation to each of the planes, all other planes play a substantially blocking role, preventing the false protection from tripping.

The high generality of the discussed method was not only its advantage, but also a disadvantage. As a matter of fact, he equalized all measurements in his rights, not giving preference to those that have already proven themselves in relay protection. The approach to the formation of a set of cells has become purely formal. As a result, a problem arose of delimiting arrays of response codes and protection blocking codes, narrowing the functionality of the method.

The purpose of the invention is to expand the functionality and at the same time simplify the method without compromising its generality. Moreover, the task is to give it an even more general form, relying on the traditional ideas of relay protection.

The goal is achieved due to the fact that the structure of any protection is reduced to modules of two types; the first generates a response signal, the second - to block the protection. Modules are trained from simulation models of an energy object in alternative modes. The new is layered structure. In contrast to the cellular structure of the prototype, where all families of cells are equal, hierarchical sequences of additional modules of both the first and second types are introduced here. The sensitivity of the first type of modules is increased as you move from the main module to the end of the sequence. Additional modules of the first type work according to the OR scheme with the main module of the same type for them. Modules of the first and second types are combined in pairs of equal hierarchy. Of fundamental importance is the fact that each module of the first type is blocked by all modules of the second type of the lower hierarchy, and in addition by a module of the same hierarchy. The created structure is being trained in alternative modes. The modes that trigger the operation of the next additional module of the first type are determined and the second type of module included in the same pair is trained to respond to these modes. The combination of the above technical features provides a phased increase in the sensitivity of the modules of the first type, and hence the entire protection, while maintaining selectivity as a function of the modules of the second type.

In the dependent claim is a special case of the proposed method, when it is not possible to present the basic protection in the form of a combination of modules of two types. In this case, the main protection becomes a single main module of the first type, and the counting of the first type of modules begins with the advent of additional modules, i.e. the first hierarchical sequence begins with the main module, and the second with the additional.

Figure 1 shows a structural diagram that implements the proposed method. Figure 2 is an illustration of the learning process of protection organized according to the proposed method. Figure 3-6 shows the illustrations necessary for the theoretical justification of the proposed method: figure 3 explains the various situations encountered when recognizing a specific mode of an energy object, figure 4 relates to the recognition of many modes, figure 5 illustrates the first (initial) stage of protection training , and Fig.6 is the second stage.

Protection 1, connected to power object 2, consists of hierarchical sequences of 3 and 4 modules of the first and second type, a combination of 5 AND elements, an end element OR 6. Three modules of each sequence are shown: modules of the first type 7-9, second type 10-12 . Modules of different types form pairs of equal hierarchy: 7 and 10, 8 and 11, 9 and 12; the lower hierarchy is in the first pair, the highest is in the third. Elements And 13-15, in turn, line up in a similar hierarchical sequence, going from 13 to 15.

The vector z is transmitted from the measuring object 2 protection 1 via the bus 16. In response to the measurement of δ z is a logical signal outputted from the output 17 element 6. The input bus 18 branches into circuit and the first circuit 19. 18 is transmitted subvector z _a, and in the second 19 - subvector z _b . Measurement z _{a is} processed by modules of the first type 7-9, and measurement z _b is processed by modules of the second type 10-12. The outputs 20-22 of the modules of the first type and the outputs 23-25 of the modules of the second type are connected to the inputs of the elements And 13-15 according to the following rule: the output of each of the modules of the first type is connected to only one of the elements And, and the output of each of the modules of the second type is connected to to all elements of an inferior and equal hierarchy. Modules 7 and 10 form a pair of the first level and are considered basic. Modules 8 and 11 form a pair of the second level of the hierarchical sequence and are at the same time the first additional modules. Modules 9 and 12 form a top-level pair if they complete sequences 3, 4. The outputs of 26-28 elements AND are independent, the element OR 6 combines them into a single output signal δ. Outputs 26-28 not only complement, but also reserve each other.

Figure 2 shows an example of retraining of protection 1, after which an even more sensitive protection 29 is obtained. The role of the teacher is a simulation model of the object 30, reproducing alternative modes. The mode is defined by the model parameter vector x _β formed by the object block 31. Protection 29 differs from protection 1 in that it contains additional modules of the first type 32 and the second type 33, which form the fourth pair. It corresponds to an additional fourth element And 34, combining the outputs of the module of the first type 32 with the outputs of all blocking modules, including the additional module 33. As a result, the And 34 element is a five-input; one input more than the previous element AND 15. The terminal element OR 35, replacing the previous element OR 6, generates an output signal 36, in which the response of the fourth channel 37, which acts along with the previous three 26, is added to the previous signal 17 -28.

In the drawings explaining the proposed method, the following notation is used:

C is the object space in which the vector of parameters x of the simulation model of the object is defined,

A is the setting space in which the measurement vector z is determined, transmitted via bus 16,

α is the general designation of the controlled modes from which the energy object 2 should be protected,

β - the general designation of alternative modes to which the protection should not respond,

With _α is the object space of α-modes,

C _β is the object space of β-modes,

And ₁ and A ₂ - setpoint spaces of the modules of the first and second type,

G is the region of the object parameters of the simulation model,

G _Σ is the domain of definition of object parameters, beyond which their values do not go beyond,

S is the area of the setting space,

F - transformation of the simulation model mode x (or the region of regimes x∈G) into measurement z (or measurement region z∈S) - direct transformation

z = F (x)

or

S = F (G),

F ^-1 - conversion of the measurement z (or measurement area z∈S) into many modes

G = F ^-1 (z)

or

G = F ^-1 (S).

The transformation F is unambiguous, and F ^-1 is usually ambiguous, since the number of object parameters usually exceeds the measurement dimension z.

Figure 3 discusses situations that arise in the study of the information properties of a particular measurement z ₁ . There are simulation models of α- and β-modes with given object regions G _αΣ and G _βΣ , direct and inverse transformations F _1α , F _1β , F _2α , F _{2β are} known; F _α1 ^-1 , F _β1 ^-1 , F _α2 ^-1 , F _β2 ^-1 .

The object region G _{βΣ is} converted to the setpoint region S _1βΣ = F _1β (G _βΣ ). On figa shows the case when the measurement of z ₁ is outside the region S _1β , therefore, in the blocking modules 4 is not necessary. The inverse α-transformation determines the region of controlled modes corresponding to a given measurement: G _α = F _α1 ^-1 (z ₁ ).

On figb depicts a more difficult situation when the measurement of z ₁ falls in the region S _1βΣ . This time, without the blocking module, recognition of emergency modes G _{α is} impossible, since the measurement of z _{1 also} corresponds to the region of alternative modes G _β = F ^-1 _β1 (z ₁ ). The blocking setting space A ₂ helps out, where the mappings of the regions G _α and G _β , namely S _2α = F _2α (G _α ) and S _2β = F _2β (G _β ), not only do not intersect, but also diverge by a sufficient distance d . By setting the response area of module 7 with coverage z ₁ and module 10 with coverage S _2β , we provide recognition of the regime G _α .

Figure 3c shows a less favorable situation when the regions S _2α and S _2β intersect in the subdomain S _2αβ , which corresponds to the object subdomain G _αβ = F _α2 ^-1 (S _2αβ ). The modes x _α ∈G _αβ turn out to be physically unrecognizable. But there remains a recognizable subdomain G _αα = G _α / G _αβ - part of the region C _α minus G _αβ . Protection of the object in the subregion G _αα becomes possible only due to the blocking of the region S _2αβ .

Finally, FIG. 3g shows the worst-case situation when measuring z ₁ does not at all make it possible to draw a conclusion about the emergency mode. This time, the region S _2α falls inside the region S _2β , and here the blocking prohibits triggering from measuring z ₁ . Thus, in this case, there is a situation of absolute unrecognizability of the α-modes displayed by measuring z ₁ .

In Fig. 4, instead of one measurement z ₁ , the measurement region S _{1 is considered} , the inverse transformations of which into object spaces of controlled and alternative modes will indicate object regions G _α and G _β indistinguishable if only one set-point space A _{1 is used} . The second setpoint space A _{2 is} shown to help. By mapping the domains G _α and G _β in this space, we obtain in the most general case the intersecting mappings S _2α and S _2β . The regions of their intersection S _2αβ corresponds to the object region G _αβ . There remains the part G _{αα of the} object domain G _α in which the α-modes meet the recognition criterion. A chain of direct and inverse transformations that ultimately determine the domain G _αα is shown in Fig. 4.

As we see, it is possible to realize the recognizing properties of the setpoint region S ₁ only if the region S _2αβ is blocked in the second setpoint space A ₂ . 5 and 6 show how this ability to recognize emergency α-modes is implemented in the proposed method. Figure 5 illustrates that often encountered situation when there is some set-point region S _1α (1) that meets the criterion of absolute recognition. This means that the map S _{1βΣ of} alternative modes does not intersect S _1α (1). The number in the argument indicates the stage of formation of the protection characteristics.

The illustrations in FIG. 6 explain the operation of two structures of FIG. 1 and FIG. 2 at once. First, we can attribute the response characteristic S _1α (1) to module 7, noting that then there is no need in module 10, and consider the characteristics of modules 8 and 11. Second, we can attribute this characteristic to the union of blocks 3 and 4 and then get the characteristics of additional modules 32 and 33. Consider the option related to the structure of figure 2, as it is focused on training protection, prior to connecting to the object 2.

At the first stage of training, blocks 3 and 4 will be formed, providing joint operation in the region S _1α (1). At the second stage, in order to increase the protection sensitivity, module 32 is introduced with characteristic S _1α (2), which is an extension of the previous characteristic S _1α (1). The new characteristic allows module 3 to operate in the subregion S _1αβ (2). The blocking module 33 is called upon to prevent false operation. Its characteristic S _2αβ (2) is built in the blocking authorized space A ₂ , for which the same modes of the simulation model 31 are displayed in the area S _1αβ (2) and in parallel in S _2αβ (2) . Blocking module 33 will affect the protection behavior not only in β-modes, for which it is intended, but also in α-modes, which is a payment for increasing sensitivity. Figure 6 shows that the module 33 with characteristic S _2αβ (2) will lead to the prohibition of the protection in the object area G _αβ (2). The region G _αα (2) remains, where the module 32 will not be blocked. Let us pay attention to a fundamentally important detail, namely, that the object response region G _α (2) of module 32 completely covers the response region G _α (1) of the protection consisting of modules 3 and 4, while the response region G _αα (2 ) the second pair of modules 32, 33 no longer includes the domain G _α (1). This fact explains the need for parallel operation of the first pair of modules 3, 4 with the outputs of the elements And 5 and the second pair 32, 33 with the output 37 of the element And 34.

Due to the delimitation of information into two groups and the display of each of them in different reference planes, the proposed method has wide functionality. Additional information should be attached to the blocking group, which will not affect the execution of traditional protection modules, but will allow to set ever more sensitive settings, blocking at the same time dangerous alternative modes.

Information sources

1. USSR Copyright Certificate No. 66343, cl. H02H 3/28, 1944.

2. RF patent No. 1775787, cl. H02H 3/40, 1991.

3. RF patent No. 2066511, cl. H02H 3/40, G01R 31/08, 1992.

4. Lyamets Yu.Ya., Nudelman G.S., Pavlov A.O. The evolution of distance protection. - Electricity, 1999, No. 3, pp. 8-15.

5. Lyamets Yu. Ya., Nudelman GS, Pavlov AO, Efimov EB, Zakonshek Ya. Recognition of damage to power transmission. Part 1, 2, 3. - Electricity, 2001, No. 2, p.16-23; No. 3, p.16-24; No. 12, pp. 9-22.

6. RF patent No. 2247456, cl. H02H 3/40, 2002.

Claims (2)

1. The method of relay protection of an energy object by constructing it from a module of the first type that generates a response signal, and from a module of the second type that generates a signal to block, and training by applying signals from simulation models of an energy object that reproduce alternative modes to controlled ones, characterized in that , in order to expand the functionality, include, according to the OR scheme with the main module of the first type, the first hierarchical sequence of additional modules of the same type, sensitivity to They increase as they move towards the end of the sequence, include in parallel with the main module of the second type a second hierarchical sequence of additional modules of the same type so that each module of the first type corresponds to one module of the second type of the same hierarchy, each module of the first type is blocked by all modules of the second type of lower and equal hierarchies, determine alternative modes that trigger the operation of the next additional module of the first type, and teach the additional module of the second type equal to hier to respond to these alternative regimens. 2. The method according to claim 1, characterized in that the main modules of the first and second types are combined and included in the first hierarchical sequence, and the second hierarchical sequence begins with an additional module of the second type.

Images