T 1635/19 (Rolling-stock maintenance/SIEMENS MOBILITY) 09-07-2021

Reasons for the Decision

1. The application relates to adaptive remote maintenance of rolling stock with the help of machine learning.

2. The request for remittal to the department of first instance for further prosecution

2.1 In its statement of grounds of appeal, the appellant requested remittal of the case to the examining division for further prosecution because "the matter of the independent claims" had not been fully considered.

In its letter of 9 June 2021, the appellant argued for the first time that the contested decision was not reasoned within the meaning of Rule 111(2) EPC, which constituted a fundamental deficiency within the meaning

of Article 11 RPBA 2007, requiring the board to remit the case to the examining division. The appellant referred to decision T 655/13, which set out the requirement that the board had to be able to examine the reasons for the impugned decision.

2.2 According to Article 11 RPBA 2020, which in view of Article 25 RPBA 2020 is applicable here, a case is not to be remitted to the examining division for further prosecution unless special reasons present themselves for doing so. As a rule, a fundamental deficiency which is apparent in the proceedings before the examining division constitutes such a special reason.

Hence, even if a fundamental deficiency were apparent, the board would still have discretion not to remit the case to the examining division but to deal with the substance of the case itself.

2.3 In the present case, there is no indication that the alleged fundamental deficiency hindered the appellant in preparing its appeal. Indeed, the appellant argued that the contested decision was not reasoned within the meaning of Rule 111(2) EPC only in response to the board's summons to oral proceedings. Nor did the alleged fundamental deficiency prevent the board from carrying out a preliminary examination of the case; the board was in fact able to examine the reasons for the impugned decision.

In view of these circumstances alone, the board considers it appropriate to reject the appellant's request for remittal of the case to the examining division for further prosecution and to continue with the examination of the merits of the case. It can therefore be left open whether the contested decision was indeed insufficiently reasoned or merely failed to convince the appellant.

Main request

3. Added subject-matter - Article 123(2) EPC

3.1 Claim 1 is directed to "a method implemented on a processing system", i.e. to a computer-implemented method. The method includes the step "issuing repair and maintenance orders for the rolling stock", which was not present in original claim 1. According to the letter of 19 November 2014 filed with the examining division, this amendment was based on paragraphs [0004] and [0073] of the published application.

3.2 Paragraph [0004] mentions repair and maintenance orders. However, this paragraph is part of the background section of the application and describes prior-art methods in which persons referred to as "fleet dispatchers" manually filter out and group together various messages and sensor data. This passage therefore does not disclose a computer-implemented step of "issuing repair and maintenance orders".

Paragraph [0073] does not mention repair and maintenance orders. It does state that "the processor 102 applies the resulting (machine-learned) statistical model or rules to the received data" and that "[a]lternatively, rules are applied by dispatchers". Even if this is understood as a reference to the "fleet dispatchers" of paragraph [0004], it does not disclose that repair and maintenance orders are issued by a processor or a processing system.

The board is not aware of any other passage in the application as filed that could serve as a basis for the amendment.

3.3 In its reply to the board's communication, the appellant did not comment on the board's objection to the feature "issuing repair and maintenance stock". In the oral proceedings before the board, the appellant confirmed that it relied on its written submissions in respect of compliance of the main request with Article 123(2) EPC.

3.4 The board therefore concludes that the subject-matter of claim 1 of the main request extends beyond the content of the application as filed, contrary to Article 123(2) EPC.

First auxiliary request

4. The first auxiliary request represents a reasonable reaction to objections raised for the first time in the board's communication and was filed at the earliest opportunity. It is therefore admitted into the appeal proceedings under Article 13(2) RPBA 2020.

5. The invention as defined by claim 1

5.1 Claim 1 is directed to a computer-implemented method ("implemented on a processing system") for adaptive remote maintenance of rolling stock. Examples of rolling stock are trains and trucks (see paragraph [0002] of the published application).

5.2 A processor receives diagnostic information from subsystems of rolling stock and identifies a plurality of events relating to maintenance incidents of rolling stock by applying "current rules" to the diagnostic information. The diagnostic information includes messages, sensor data, or messages and sensor data.

5.3 The processor further receives information "corresponding to event validation and event discarding of the events relative to the respective current rules".

The board notes that this information may have been generated by a human operator who evaluates the identified events and discards those which are not relevant (paragraphs [0031] and [0032]).

5.4 The processor generates a new rule different from the current rules by applying supervised learning to the diagnostic information with the information corresponding to event validation and event discarding as ground truth information.

5.5 The new rule is output and applied.

6. Inventive step

6.1 Document D5 relates to a computer-implemented system and method for monitoring the health of a fleet of assets such as aircraft, ground-based vehicles and locomotives and for remotely maintaining the fleet (paragraphs [0001] and [0014]). The health of an asset is monitored by instrumenting the asset with sensors and monitoring sensor readings to detect signs of failure (paragraph [0002]).

Monitoring sensor readings involves comparing sensor values related to an asset with parameters that are estimated on the basis of a model of the asset (paragraph [0015]). The model is initially generated on the basis of historical data (ibid.). The parameters of the model are re-determined on the fly on the basis of current observations of data values (paragraph [0017]). Alerts are generated when sensor values deviate from estimated values by more than a threshold value (paragraphs [0037] and [0038]).

The sensor values and alerts, i.e. diagnostic information within the meaning of claim 1, are provided to an "incident diagnostics engine", which applies rules to the diagnostic information to generate a plurality of events referred to as "incidents" (paragraph [0039]). These incidents are added to a watch list (ibid.).

Incidents on the watch list are presented by means of a graphical user interface (paragraph [0158]), thereby allowing human experts to take appropriate actions (paragraph [0013], last sentence, and paragraph [0014]; paragraph [0007]).

6.2 Hence, the method of claim 1 differs from the disclosure of document D5 in that it comprises:

(a) receiving "information corresponding to event validation and event discarding of the events relative to the respective current rules";

(b) applying supervised learning to the diagnostic information with the "event validation and event discard" as ground truth information;

(c) outputting a new rule from the applying of the supervised learning, the new rule being different than any of the current rules;

(d) applying the new rule.

In addition, claim 1 specifies that the diagnostic information is received from subsystems of rolling stock (distinguishing feature (e)).

6.3 The distinguishing feature (e) is obvious given that assets of rolling stock typically comprise subsystems having their own sensors (see published application, paragraph [0003]; D5, paragraph [0003]).

6.4 The distinguishing features (a) to (c) express that a new rule for the identification of events/incidents from diagnostic information is created by applying supervised learning to the diagnostic information while using the information indicating which of the generated events/incidents have been validated and which have been discarded by human experts as ground truth information. Feature (d) then expresses that the newly generated rule is applied (to future diagnostic information).

6.5 The distinguishing features (a) to (d) relate to an abstract scheme of learning a new classification rule (for classifying certain combinations of sensor values as events/incidents) on the basis of input data (the diagnostic information) and expected output data (the "ground truth information" indicating whether the events generated by the current rules are correct (by having been validated) or not (by having been discarded)). This abstract scheme is non-technical "as such". This means that the scheme is to be taken into account when assessing inventive step only to the extent that it interacts with the technical subject-matter of the claim to solve a technical problem (T 154/04, OJ EPO 2008, 46, Reasons 5, under (F)).

6.6 The appellant argued that, since the claim included features specifying that the steps of the method were implemented on a processing system and performed "by a processor", all the non-technical features of the claim contributed to solving a technical problem.

However, the use of technical means to carry out a non-technical feature does not mean that the non-technical feature interacts with the technical means to solve a technical problem. Rather, the technical means are used to solve the problem of implementing the non-technical feature. Since in the case at hand the technical means consist in a processing system and a processor, both being well-known and used for their normal purpose, no inventive contribution can be seen in the implementation of the non-technical abstract scheme.

6.7 The appellant also argued that all claim features had to be assessed individually when assessing their technical contribution and that "lumping them together", as both the examining division and the board had done, was incorrect.

However, beyond the trivial and thus obvious aspect of their implementation on a computer, the board cannot see how any of the individual features of claim 1, when considered in isolation, could make a technical contribution. It is therefore the combination of the claimed features that requires further consideration.

6.8 The appellant further argued that the rules were used to filter diagnostic information including sensor data. By improving the rules by means of a learning routine, rolling stock issues could be identified with better accuracy. As those issues were measured by sensors, they were of a technical nature. More accurate rules therefore contributed to solving a technical problem.

However, the quality of the new rule in terms of its effect on the accuracy of the identified events fully depends on the quality of the ground truth information. The origin of this information is not specified in the claim, and the description confirms that it may be the result of cognitive processing by a human operator dealing with the events. Any improvement in the quality of the event identification therefore is not a technical effect achieved by the features of the claim. In addition, the accuracy of the identified events is not a technical parameter, since it is judged by the human operator.

6.9 In view of the above, the board judges that the distinguishing features (a) to (d) do not provide a technical contribution going beyond their straightforward and thus obvious implementation using well-known technical means.

6.10 Hence, the subject-matter of claim 1 of the first auxiliary request lacks inventive step (Article 56 EPC).

Conclusion

7. Since none of the appellant's requests is allowable, the appeal is to be dismissed.