DESCRIPTION

The present invention relates to a method for reading a digital graphical diagram representing an electric circuit, for example an electric power distribution grid.

As is known, an electric circuit, such as an electric power distribution grid, may be graphically described in digital format using different semantics for representing the electrical components and the electrical connections of the electric circuit.

In general, information content and graphical symbols adopted in a graphical diagram depend on the software program used to design the electric circuit and/or on the personal choice of the circuit designer.

Examples of different digital graphical diagrams representing electric circuits are schematically shown in figures 1-3. It is apparent how each graphical diagram includes different information content and adopts a different semantics to represent components and their electrical connections.

Graphical diagrams related to electric circuits need often to be combined, updated or further processed, e.g. for the purpose of generating a BOM (Bill of Materials) or preparing a commercial offer related to an electric power distribution plant to be installed.

Nowadays, due to the available variety of information details and semantics to describe electric circuits, all these activities generally entail the intervention of highly specialized and skilled personnel and are time-consuming/expensive to carry out.

The main aim of the present invention is to provide a method for reading a digital graphical diagram representing an electric circuit, which allows solving or mitigating the technical problems evidenced above.

Within this aim, an object of the present invention is to provide a method that allows collecting information about the structure of an electric circuit represented in a digital graphical diagram in a quick and effective way.

A further object of the present invention is to provide a method that can be easily carried out even by non- specialized operators.

A further object of the present invention is to provide a method that can be easily computer- implemented without the adoption of expensive processing resources.

These aim and objects are achieved by a method for reading a digital graphical diagram representing an electric circuit, according to the following claim 1 and the related dependent claims. In a general definition, the method, according to the invention, comprises the following steps: for each diagram page, detecting the graphical objects included in said diagram page; for each diagram page, basing on the detected graphical objects, obtaining predictive information related to the components included in the portion of electric circuit represented in said diagram page;

for each diagram page, harmonising predictive information related to the components of the portion of electric circuit represented in said diagram page and providing an identification list of the components said electric circuit.

According to an aspect of the invention, the step of detecting the graphical objects included in each said diagram page includes obtaining identification data related to the detected graphical objects. Said identification data conveniently include predictive information related to the class, the position and the size of the detected graphical objects.

According to an aspect of the invention, the step of obtaining predictive information related to the components of the portion of electric circuit represented in said diagram page includes merging and harmonising the identification data related to the detected graphical objects of said diagram page.

According to an aspect of the invention, the step of obtaining predictive information related to the components of the portion of electric circuit represented in said diagram page includes detecting one or more graphical connectors among detected component objects of said diagram page.

According to an aspect of the invention, the step of obtaining predictive information related to the components of the portion of electric circuit represented in said diagram page includes assigning one or more detected text objects or table objects of said diagram page to one or more detected component objects of said diagram page.

According to an aspect of the invention, the method comprises a step of obtaining, for each diagram page, one or more first crop portions of said diagram page. Each first crop portion includes partially overlapped regions of the corresponding diagram page.

According to an aspect of the invention, the method comprises a step of obtaining, for each diagram page, additional predictive information related to the components of the portion of electric circuit represented in said diagram page.

Such a step of obtaining additional predictive information includes:

for each diagram page, obtaining one or more second crop portions of said diagram page, said second crop portions being formed by text areas or table areas of said graphical diagram at said diagram page; for each diagram page, detecting text strings included in the second crop portions of said diagram page;

for each diagram page, basing on the detected text strings, obtaining predictive information related to assigning one or more detected text objects or table objects included in said diagram page to one or more detected component objects of said diagram page.

According to an aspect of the invention, the method comprises a step of pre-processing a graphical file including said graphical diagram in digital format.

Such a pre-processing step includes:

converting said graphical file into a first image file including said graphical diagram in digital format;

splitting said first image file into one or more second image files, each including a diagram page in digital format;

if necessary, reducing the size of said second image files.

Preferably, the above-mentioned step of obtaining the first crop portions of a diagram page includes splitting a second image file related to said diagram page into a batch of third image files, each including a first crop portion said diagram page in digital format.

Preferably, the above-mentioned step of obtaining the second crop portions of a diagram page includes splitting a second image file related to said diagram page into a batch of fourth image files, each including a second crop portion of said diagram page in digital format.

According to an aspect of the invention, the method comprises a step of extracting, from the above-mentioned identification list, information about the properties of the components included of said electric circuit.

The method, according to the invention, is apparently intended for a computer- implementation. In some further aspects, the present invention thus relates to a computer program, according to the following claim 13, and to a computerised device or platform, according to the following claim 14.

Further characteristics and advantages of the present invention will emerge more clearly from the description of preferred, but not exclusive embodiments, of which non-limiting examples are shown in the attached drawings, in which:

figures 1-3 are schematic views showing different types of digital graphical diagrams representing electric circuits;

figures 4-8 are schematic views showing the steps of the method, according to the invention. With reference to the above-mentioned figures, the present invention refers to a method 1 for reading a graphical diagram 100 in digital format, which represents an electric circuit.

In principle, the above-mentioned electric circuit may be of any type.

The method, according to the invention, is however particularly adapted for reading graphical diagrams related to extended electric power distribution grids.

The graphical diagram 100 includes one or more diagram pages, each representing a portion of the electric circuit represented.

An example of graphical diagram is shown in figure 4. In this case, the graphical diagram 100 includes a single page only.

In a practical applications, however, the graphical diagram 100 may have a huge number (e.g. some tens) of diagram pages.

In general, the graphical diagram 100 may adopt a variety of graphical symbols of different types to represent components of an electric circuit and their electrical connections.

However, regardless of the semantics used to represent said components and electrical connections, each diagram page of the graphical diagram 100 may include one or more graphical objects and one or more graphical connectors (figure 4).

Graphical objects of the graphical diagram 100 generally relate to components of the electric circuit, text areas or table areas of said the graphical diagram. The graphical diagram 100 thus includes one or more of the following graphical objects: component objects, text objects and table objects.

Component objects of the graphical diagram 100 generally represent corresponding components of the represented electric circuit (e.g. an electric load, a circuit breaker a transformer, and the like).

Text objects of the graphical diagram 100 generally represent corresponding text areas of said graphical diagram.

Table objects of the graphical diagram 100 generally represent corresponding text areas of said graphical diagram.

All these graphical objects may have different size and configuration depending on the type of the circuit components and on the semantics used to graphically represent said components. Graphical connectors of the graphical diagram 100 generally relate to electrical connections among two or more components of the represented electric circuit.

Preferably, each graphical connector is represented as a single line linking at least two component objects of the graphical diagram.

The steps of the method 1, according to the invention, will be now described in details with reference to its implementation by a computerised device or platform.

According to a preferred embodiment of the invention, the method 1 comprises a preliminary step 17 of pre-processing a graphical file including the graphical diagram 100 in digital format.

Conveniently, the graphical file may of the“.pdf’ or“.dxg” type or other similar graphical format.

Preferably, the preliminary step 17 of the method 1 includes a sub-step of converting the above-mentioned graphical file into a first image file, e.g. of the“.jpg” or“.png” type or other similar image format. In this way, the above-mentioned graphical file is converted into a readable matrix of pixels, which is more adapted for being further processed.

Preferably, the preliminary step 17 of the method 1 includes a sub-step of splitting the above- mentioned first image file into one or more second image files, each including a diagram page of the graphical diagram 100.

This solution allows remarkably reducing the computational load required to carry out the following main steps of method 1.

In order to further reduce such a computational load, the preliminary step 17 of the method 1 preferably includes a sub-step of reducing, if necessary, the size of one or more second image files obtained at the previous splitting step.

Conveniently, image files representing diagram pages of the graphical diagram 100, which are too huge for being processed in a reasonable time, are resized to maximum predefined dimensions.

According to a preferred embodiment of the invention, the method 1 comprises a step 11 of obtaining, for each diagram page of the graphical diagram 100, one or more first crop portions of said diagram page (figure 5).

Conveniently, the first crop portions obtained for each diagram page are selected in such a way to include partially overlapped regions of said diagram page. In this way, the probability of missing graphical information located at the borders of the first crop portions is remarkably reduced.

Obviously, as shown in figure 5, the first crop portions obtained for each diagram page are tailored in such a way that they cover the whole image represented in a diagram page.

In a practical implementation of the method 1, the first crop portions of each diagram page are obtained by cropping each second image file related to said diagram page. In this way, a batch of third image files, each including a first crop portion in digital format, is obtained.

The obtaining of the above-mentioned first crop portions is basically directed to dramatically reduce the computational load required for detecting graphical objects and the graphical connectors in each diagram page of the graphical diagram 100.

The computational load related to the detection process of the graphical objects included in each diagram page, and more generally in the graphical diagram 100, can in fact be parallelised among a number of smaller images, which can be processed more easily.

According to the invention, the method 1 comprises a step 12 of detecting, for each diagram page of the graphical diagram 100, the graphical objects included in said diagram page, preferably in the first crop portions of said diagram page (figure 6).

As shown in figure 6, detected graphical objects may be component objects, text objects or table objects.

Preferably, the detection step 12 includes obtaining identification data related to the repeatedly detected graphical objects of each first crop portion.

Preferably, these identification data include predictive information related to the class (e.g. the type) of each detected graphical object, the position of each detected graphical object and the size of each detected graphical object.

Referring to figure 6, practical examples of said identification data may be the following:

{detected graphical object #1: class (table area), position (xl, yl), size (SI)};

{detected graphical object #2: class (circuit breaker), position (x2, y2), size (S2) } ;

{detected graphical object #3: class (text area), position (x3, y3), size (S3)}.

In a practical implementation of the method 1, the detection step 12 may be carried out by exploiting suitable TM (Template Matching) algorithms, which may be of known type. As an alternative, expert systems employing suitably trained neural networks may be used.

According to the invention, the method 1 comprises a step 13 of obtaining, for each diagram page of the graphical diagram 100, basing on the detected graphical objects, predictive information related to the components included in the portion of electric circuit represented in said diagram page.

In one aspect, the step 13 of the method 1 is directed to assemble and optimize the predictive information collected at the previous detection step 12 of the method 1.

Preferably, the step 13 of the method 1 comprises a sub-step of merging and harmonising, for each diagram page of the graphical diagram 100, the identification data related to the detected graphical objects of said diagram page.

Preferably, such a merge and harmonisation step includes converting the position coordinates of the detected graphical objects (which are normally referred to the reference system of the respective first crop portion) into absolute position coordinates referred to the reference system of the diagram page.

In a practical implementation of the method 1, such a conversion process may be easily carried out by exploiting suitable conversion algorithms, which may be of known type.

Preferably, such a merge and harmonisation step includes deleting duplicated identification data related to identical detected graphical objects of each diagram page. Such a filtering process is made necessary by the circumstance that a same graphical object may be detected many times, e.g. at adjacent overlapped first crop portions.

In practice, for each diagram page of the graphical diagram 100, the predictive information resulting from such a filtering process will be given by the difference between the union and the intersection of the predictive information included in the data related to the detected graphical objects in said diagram page.

In another aspect, the step 13 of the method 1 is directed to integrate the predictive information collected at the previous detection step 12 of the method 1.

Preferably, the step 13 of the method 1 comprises a sub-step of detecting one or more graphical connectors among two or more detected component objects included in said diagram page.

In this way, for each diagram page, it is possible to collect predictive information on how the components of the portion of electric circuit represented in said diagram page are operatively connected.

In a practical implementation of the method 1, such a detection process may be carried out through suitable feature extraction algorithms of known type, such as LSD (Line Segment Detection) algorithms, HT (Hough Transform) algorithms or DSCC (Directional Singly Connected Chain) algorithms. As an alternative, expert systems employing suitably trained neural networks may be used.

In yet another aspect, the step 13 of the method 1 is directed to structure the predictive information collected at the previous detection step 12 of the method 1 in such a way to reconstruct the whole information content of each diagram page.

Preferably, the step 13 of the method 1 comprises a sub-step of assigning, for each diagram page, one or more detected text objects or table objects included in said diagram page to one or more detected component objects included in said diagram page.

Such an assignment process may be carried out according to a variety of different criteria.

As an example, proximity criteria may be used: a detected text object or a table object of diagram page (representing a text area or a table area of the diagram page) will thus be assigned to the nearest component object of said diagram page (representing a component of the electric circuit portion represented in the diagram page).

It is evident how the above-described assignment process is directed to associate, for each diagram page, each text area or table area of said diagram page to the correct component of the electric circuit portion represented in said diagram page.

According to a preferred embodiment of the invention, the method 1 comprises a step 16 of obtaining, for each diagram page, additional predictive information related to the components of the portion of electric circuit represented in said diagram page.

The step 16 of the method 1 is conveniently directed to recover additional information contents about the components of the portion of electric circuit represented in each diagram page. Such additional information contents may be used to improve the quality of the predictive information collected at the previously-described steps 12-13 of the method 1. Preferably, the step 16 includes a sub-step of obtaining, for each diagram page, one or more second crop portions of said diagram page, which are formed by text areas or table areas of the graphical diagram 100 at said diagram page (figure 7).

Conveniently, the second crop portions obtained for each diagram page are tailored in such a way to include only text areas or table areas of the graphical diagram 100.

In practical implementation of the method 1, the second crop portions of each diagram page are obtained by cropping each second image file related to said diagram page. In this way, for each diagram page, a batch of fourth image files, each including a second crop portion in digital format, is obtained.

For each diagram page, such a batch of fourth image files can thus be easily processed to extract the information content of each text area or table area of the graphical diagram 100. Preferably, the step 16 includes a sub-step of detecting, for each diagram page, the text strings included in each second crop portion of said diagram page.

In a practical implementation of the method 1, such a detection process may be carried out through suitable OCR (Optical Character Recognition) algorithms of known type.

Preferably, the step 16 includes a sub-step of obtaining, for each diagram page, basing on the detected text strings obtained through the above-mentioned detection process, predictive information related to the assignment of a detected text object or table object of said diagram page to a detected component object included in said diagram page.

In a practical implementation of the method 1, such a string analysis process may be easily carried out using suitable probabilistic classifiers (e.g. Naive-Bayes classifiers) or probabilistic decision networks (e.g. Bayesan networks), which may be of known type.

As an alternative known RegEx (Regular Expressions) extraction algorithms suitable to extract information from the detected text strings may be used.

The additional predictive information collected at the step 16 of the method 1 allows improving the assignment of a detected text object or table object of each diagram page to a detected component object included in said diagram page.

A detected text object or a table object of diagram page may be assigned to a component object of said diagram page basing on the information contained in said text object or table object and extracted from the text strings detected in said text object or table object.

According to the invention, the method 1 comprises a step 14 of harmonising, for each diagram page, the predictive information related to the components of the portion of electric circuit represented in said diagram page.

Predictive information related to the components of the portion of electric circuit represented in said diagram page, which has been collected at the previously described steps 12, 13 and, possibly, 16, of the method 1, may in fact be contradictory or inhomogeneous.

The step 14 of the method 1 is thus directed to filter the collected pieces of information in order to eliminate or reduce possible discrepancies.

In a practical implementation of the method 1, such a string analysis process may be carried out basing suitable known selection algorithms based on predefined rules, e.g. upon assignment a priority level to each collected piece of information.

As an alternative, Machine Learning or Graph Network algorithms of known type may be conveniently used.

Conveniently, the step 14 of the method 1 includes providing an identification list of the components of the electric circuit represented in the graphical diagram 100. Such an identification list conveniently includes also information about the related electrical connections and about the text areas or table areas assigned to said components.

According to a preferred embodiment of the invention, the method 1 comprises a step 15 of extracting from the above-mentioned identification list information about the properties of the components included of said electric circuit.

Conveniently, such an extraction steps may be carried out by analysing the text strings associated to each component of the electric circuit. To this aim, known RegEx (Regular Expressions) extraction algorithms may be used.

It is apparent from the above how the step 15 of the method 1 is directed to improve the quality of information derivable from the graphical diagram 100 by automatically inferring information that is implicitly contained in text areas or table areas of the graphical diagram itself. The method 1, according to the invention, is particularly suitable for being implemented by a computerised device or platform.

In a further aspect, the present invention thus relates to a computerised device or platform is provided with processing resources capable of executing software instructions stored or storable in a storage medium to implement the method, according to the invention.

Such a computerised device or platform may be of any type, e.g. a desktop or portable computerised device, a server platform, a cloud-computing platform, and the like.

In yet a further aspect, the present invention thus relates to a computer program including software instructions stored or storable in a storage medium to implement the method, according to the invention, when executed by a computerised device or platform.

The method 1, according to the invention, allows fully satisfying the above-mentioned aim and objects.

The method allows collecting information about the structure of an electric circuit represented in a digital graphical diagram in a quick and effective way, regardless of the semantics used for representing the electrical components and the electrical connections of said electric circuit.

Graphical diagrams representing electric circuits may thus be analysed uniformly and with a high level of accuracy, independently of the specific skills or experience of the operator.

The method is particularly suitable for implementation at industrial level by means of a computerised device or platform without requiring particularly powerful and expensive data processing resources.