T 0124/22 (Cloud-modelling of an automation system/ROCKWELL) 19-07-2024

Reasons for the Decision

1. Procedural matters

1.1 Late notification of non-attendance

In this case, the respondent's representative provided his videoconferencing details eight days before the oral proceedings, indicating an intention to participate. However, he notified the board of his

non-participation only one day before the scheduled proceedings (cf. point IV above). Typically, such notifications are given well in advance (see also T 930/92, Headnote I). Given that the board's preliminary opinion was issued ten months ahead of the scheduled hearing, the respondent had ample time to inform the board of its non-attendance well ahead of the hearing.

While it is not uncommon for representatives to receive late instructions, they should seek timely directions from their clients, particularly when arranged oral proceedings approach. In this instance, the representative failed to communicate promptly with the board's registry. Instead, the board received a brief written notice only one day before the arranged oral proceedings, without any further explanations. The board (and presumably the opposing party's representative) had already invested some time in preparing for the oral proceedings. According to Article 6 of the epi Code of Conduct, members are required to act courteously in their dealings with the EPO. The same principle applies to behaviour towards other representatives (see Article 5(a) of the epi Code of Conduct).

1.2 Decision in written proceedings

The respondent effectively withdrew its request for oral proceedings by declaring its intent not to attend them. In turn, the board did not consider the conduct of oral proceedings to be expedient (cf. Article 116(1) EPC). As a consequence, the decision is handed down in written proceedings (Article 12(8) RPBA).

2. Technical background

2.1 The opposed patent relates to the modelling of an industrial automation environment. It addresses the problem of providing an improved system for generating, on a "cloud platform", a model of an industrial automation system that corresponds to the real system based on data from its devices. The solution involves a "modeller component" that generates a model. According to the opposed patent, this model can be, for instance, a multi-dimensional model, facilitating a

multi-dimensional (e.g. "3D") view of the system. Moreover, this model can be used to interact with the real system, enabling remote viewing, remotely controlling the system's operations and remote troubleshooting.

2.2 Figure 1 (reproduced below) of the opposed patent shows a high-level overview of the system's architecture and its interaction with an industrial automation environment.

FORMULA/TABLE/GRAPHIC

In particular, Figure 1 illustrates industrial automation system 106 which represents the real-world system being modelled. Furthermore, industrial automation system 106 comprises various industrial devices 110 from which data is collected. This collected data is stored in data store 118 and used by modeller component 102 to generate the model of the industrial automation system based on the collected data.

3. Main request: claim 1 - construction

3.1 In point 4.1 of the statement of grounds of appeal, the appellant re-iterated its arguments summarised in Reasons 3.1 of the appealed decision regarding claim construction. The board considers it to be expedient to focus in this respect on the terms

- "data" according to features (b) and (c) (see point 3.2 below)

and

- "model" in accordance with feature (e) (see point 3.3 below).

3.2 Reasons 3.2.2 of the appealed decision states that "the term ['data'] is nowadays common general knowledge for communications and computing or computer assisted systems". The board does not question this but notes that features (b) and (c) use the term "data" in several contexts (emphasis added):

- "a set of industrial data from a set of devices";

- "respective data [that] is filtered by a segmenter component";

- "data that is to be provided to the modeler component";

- "data that is not to be provided to the modeler component";

- "the data [that is] associated with the industrial automation system".

While the skilled reader may regard the first and fifth options above to relate, as a first group, to the same "data" and likewise for the second to fourth options above as a second group, they would readily understand these two groups not to necessarily concern the same "data": the fact that the filtering in accordance with feature (c) of the "data" of the second group is "based on respective levels of data sensitivity" of the "data" of the first group does not require the "data" of the two groups to be the same.

Moreover, as correctly pointed out by the appellant, the "segmenter component" of feature (c) is, contrary to what is disclosed in paragraph [00133] of the original application, not mandatorily part of the claimed system. It follows that the claimed system only needs to concern the "data" of the first and fifth options above, i.e. of the "set of industrial data" according to feature (b). Granted claim 1 merely poses the following two restrictions on this "set of industrial data":

- it must be storable in a data store in accordance with feature (d);

- it must be possible to associate "respective levels of data sensitivity" with this data, as set out in feature (c).

3.3 Reasons 3.2.1 of the appealed decision opines that "the person skilled in the art does well know how the term ['modeling'] is to be understood". Again, the board sees no reason to question this. In particular, the skilled reader would, based on their common general knowledge, be aware of several model types that are frequently used in science and engineering, such as:

1) a visual model type, like a flow diagram for

processes in a petrochemical plant;

2) a mathematical model type, like the general

theory of relativity used in GPS-guided vehicles

that interface with material-handling robots

(cf. paragraph [0125] of the opposed patent);

3) a computational model type, e.g. to simulate

1/f noise in an electrical circuit with a

computer system;

4) a physical model type, like a building's scale

model made by a 3D-printing system

and

5) an operational model type, involving concrete,

replicable procedures for representing a

particular phenomenon in a physical system, such

as to measure the mechanical hardness of a

material produced by an industrial process or to

optimally position a sonotrode in a "smart

welder" (cf. paragraph [0124] of the opposed

patent).

According to feature (e), the "modeler component" mentioned in feature (a) must generate "a model of the industrial automation system". Besides that, it must be generated based on the "set of industrial data" obtained by the mesh network component "from a set of devices of an industrial automation system" in accordance with feature (b), granted claim 1 specifies no details regarding this model. For instance, no details are specified regarding

- the model's input or output,

- the accuracy and technical relevance of outcomes, insights or predictions provided by the model

and

- the circumstances under which the model is used.

Given this lack of details, the board holds that the "model of the industrial automation system" can be of any of the model types 1) to 5) mentioned above. As is apparent from the examples cited for these five model types above, each of these five model types can be readily applied in an "industrial automation system". Moreover, the "data" normally underlying these five model types can very well, depending on the circumstances, satisfy the two restrictions mentioned in the last paragraph of point 3.2 above. For instance, each of those five model types can in principle rely on digital data that are under copyright or another form of data protection. Furthermore, claim 1 as granted does not provide any details on the "result" or on the "analysis" of the set of industrial data in accordance with feature (e) either. This "analysis" could, for instance, merely relate to the amount of data that is provided, to the type of data (e.g. "video" or "audio") or to the integrity of the data. Accordingly, the "result" of this "analysis" could be a number, a type, a bit-error report or even a plain acknowledgement that the analysis was successful or inconclusive.

4. Main request: claim 1 - sufficiency of disclosure

4.1 Reasons 3.2.1 of the appealed decision suggests that the skilled person would in fact know how to generate the "model" in accordance with feature (e) based on paragraph [0013] of the opposed patent or even simply using their common general knowledge. The skilled person would allegedly know how to do so "independently from the chosen structure/configuration of such 'modeling' activity". However, the appellant disputes that paragraph [0013] of the opposed patent provides sufficient details in this respect.

4.2 The board notes that paragraphs [0103] to [0105] of the opposed patent state that a "simulation model" of an industrial automation system is generated based on "the set of data relating to the industrial automation system 206". It agrees in this regard with the appellant that the "person skilled in the art is not able to identify, in the patent as a whole, which set of data is meant" in features (d) and (e). This is because, as further explained in points 4.2.1 to 4.2.5 below, the opposed patent provides only limited information on the "set of data" to be used to generate the "simulation model":

4.2.1 These paragraphs [0103] to [0105] at most disclose the "set of data" and the associated analysis to be such that the "properties, characteristics, functions, etc." of building blocks of industrial automation system 206 (e.g. blocks 210, 212, 214 and 216 shown in Figure 2 of the opposed patent) or their "interrelationships" can be simulated or emulated.

4.2.2 Paragraph [0064] of the opposed patent suggests that the "data" can relate, for instance, to the type or version of the software utilised by the industrial automation system. However, it does not set out how this kind of data is actually used to generate the "model" according to feature (e).

4.2.3 Paragraphs [0075] and [0076] of the opposed patent describe a scenario where "modeler component 202" analyses "the data" and updates model 204 to reflect or represent a "motor short" that may cause a "fluid spill" in industrial automation system 206. Updated model 204 can then be used by a virtualisation component 224 to update virtualised industrial automation system 226 and present a graphical representation illustrating the motor short and fluid spill. Paragraph [0013] of the opposed patent refers in this respect also to "relationships" between "constituent components" of the industrial automation system and "a multi-dimensional view (e.g. 3-D view or 2-D view) of the industrial automation system" (emphasis added).

4.2.4 Moreover, paragraph [0081] of the opposed patent seems to suggest that the "data" can relate to the role that certain "users" can assume with respect to the industrial automation system, such as a "maintenance engineer", a "shift supervisor" or a "network engineer". It is, however, not explained how these user roles take part in generating the "model" in accordance with feature (e).

4.2.5 Likewise, paragraph [0176] of the opposed patent mentions "extrinsic data" such as data relating to "energy cost", "material cost" or "web site traffic statistics". Yet, it is also not explained how this data contributes to generating the "model" in accordance with feature (e).

4.2.6 Therefore, the board considers that the opposed patent at most concerns a "simulation model" that can be used to make a (multi-dimensional) graphical representation.

4.3 This in turn means that, of the five model types mentioned in point 3.3 above, the opposed patent concerns only model type 3) and to some extent also model type 1). How the "modeler component" of feature (e) could generate a "model" of, for instance, model types 2) and 5) is however not disclosed in the opposed patent. This would, contrary to what was found in Reasons 3.2.1 of the appealed decision (cf. point 4.1 above), also not be apparent based solely on a skilled person's common general knowledge, given that model types 2) and 5) would, at least in general, require extensive research, knowledge and expertise. The "big data" analysis mentioned in, for instance, paragraphs [0020], [0029] and [0129] of the opposed patent may provide some assistance in this respect, but, in the board's view, it will typically require a highly-trained specialist to manage the relevant data. Moreover, myriads of platforms are available for analysing "big data", typically offering a different trade-off between accuracy, usefulness and cost. Which one to take for generating the "model" in accordance with feature (e) will depend, for instance, on the model's intended purpose, e.g. predicting a system's behaviour, developing a new system or optimising an existing one. The opposed patent only provides vague statements in this respect, such as

- "provide an improved system" in paragraph [0009],

- "facilitate interacting with (e.g., remotely monitoring operation of, tracking operation of, controlling operation of, troubleshooting problems with, providing assistance relating to, etc., via a communication device) the industrial automation system" in paragraph [0029],

- "simulating or emulating the properties, characteristics, functions, etc., of the respective devices, processes, and/or assets of the industrial automation system" in paragraph [0103]

and

- "cloud platform 502 [hosting the modeler component as per feature (a)] can be any infrastructure that can allow cloud services" (emphasis added) in paragraph [0127].

4.4 Moreover, even when confining the model types to merely model types 1) and 3), the invention defined by granted claim 1 is not sufficiently disclosed by the opposed patent, at least not for all technically meaningful interpretations which would objectively occur to the skilled reader based on their common general knowledge (cf. T 149/21, Reasons 3.6). This is because it would not be apparent for the skilled person, neither based on their common general knowledge nor based on the teaching of the opposed patent, how to actually carry out the invention defined by claim 1 for the examples mentioned in the last paragraph of point 3.3 above, i.e. for those sets of industrial data, analyses of these sets and results of these analyses that are not restricted to "properties, characteristics, functions, etc." of selected building blocks of the industrial automation system or their "interrelationships" (cf. point 4.2.1 above).

4.5 Hence, contrary to the opposition division's conclusion drawn in Reasons 3 of the appealed decision, the opposed patent does not disclose the invention defined in claim 1 of the main request in a manner sufficiently clear and complete for it to be carried out by the skilled person. As a consequence, the ground for opposition under Article 100(b) EPC prejudices the maintenance of the patent as granted.

5. Auxiliary requests: claim 1 - sufficiency of disclosure

5.1 The appealed decision was not, within the meaning of Article 12(2) RPBA, "based on" the fifteen auxiliary requests mentioned in point 3 above since the opposition division rejected the opposition. In such a case, the proprietor has to demonstrate that these requests were "admissibly filed and maintained" in the proceedings before the opposition division (cf. Article 12(4), first sentence, RPBA). If this is done, the auxiliary requests form part of the appeal proceedings.

5.2 In the case in hand, the board notes that, irrespective of this admittance issue, the invention as defined in claim 1 of any of these fifteen auxiliary requests is not sufficiently disclosed either. This is because none of features (f) to (z) provides a remedy for at least the deficiency mentioned in point 4.4 above. In particular, the "model management component" and "inventory" mentioned in features (q) and (s) do not concern how to generate a "model" in accordance with feature (e) based on any of the "sets of industrial data", the analyses of these sets and the results of these analyses set out in the last paragraph of point 3.3 above. Likewise, the "set of legacy industrial device data" as per feature (w) can still be analysed merely in relation to the amount, the type or the integrity of the data, as explained in the last paragraph of point 3.3 above.

5.3 Hence, the fifteen auxiliary requests are not allowable under Article 83 EPC either.