FR3127055A1 - Electronic decision support device for the implementation of a critical function or an assistance function by an avionic system, associated method and computer program - Google Patents

Abstract

Electronic decision support device for the implementation of a critical function or an assistance function by an avionic system, method and associated computer program The invention relates to an electronic decision support device. decision (24) for the implementation of a critical function or an assistance function by an avionics system (10) in response to a request transmitted by an operator (14). The device comprises a module for receiving the request and a current context and a processing module configured to generate a recommendation in response to the request and comprising: + a similarity reasoning module configured to generate a recommendation from an algorithm based on the contents of a reference database; + a rule-based reasoning module configured to generate a recommendation from a deterministic algorithm; + an ontology reasoning module configured to generate a recommendation from an ontology-based algorithm; + an activation module configured to activate at least one of the reasoning modules. Figure for the abstract: Figure 1

Description

Electronic decision support device for the implementation of a critical function or an assistance function by an avionic system, associated method and computer program

The present invention relates to an electronic decision support device for the implementation of a critical function or an assistance function by an avionic system.

The invention also relates to a control station comprising such an electronic decision-making aid device.

The invention also relates to a decision support method for the implementation of a critical function or an assistance function by an avionic system.

The invention also relates to a computer program comprising software instructions which, when executed by a computer, implement such a method.

The avionics system is in particular embedded in an aircraft or in a remote control station of an aircraft or else in a control station of the aircraft.

The invention relates to the implementation of critical avionic functions, that is to say which are critical for the safety of the associated aircraft, the operators, the passengers and/or the environment of the aircraft.

The invention also relates to the implementation of so-called assistance functions, that is to say which make it possible to facilitate the operations carried out by the operator or to increase the level of information of the operator. These avionic assistance functions, with a lower level of criticality than the avionic functions, can be physically embedded on EFB (Electronic Flight Bag) or equivalent type computers, on mission systems, on tablets, in a cloud, in a ground support tool of the operator or air traffic control authorities.

Flight controls or communication with the exterior of the aircraft are examples of such critical functions. Functions for displaying recommended diversion airports on a sound card, for example assistance functions.

In particular, in the avionics field, a critical function is for example defined by the ARP-4754A (Aerospace Recommended Practice) standard.

By implementation of a critical function, we mean the performance of one or more calculations making it possible to generate at least one output datum associated with this critical function, from at least one input datum. The implementation of an assistance function means the performance of one or more calculations making it possible to generate at least one output datum associated with this assistance function, from at least one datum of entrance.

The invention relates in particular to the general field of assistance to an operator of the avionics system making it possible to resolve complex requests from the operator in order to help him in his decision-making. A request is in particular a search for information or a solution expressed by the operator. In particular, “complex request” means a request involving complex calculations, relating to a large quantity of data to be processed and with multifactorial constraints.

We know assistance devices based on an artificial intelligence algorithm to solve these complex requests.

The assistance device uses, for example, an expert system. Such an expert system uses a large base of formal rules and is used by the operator in the form of successive questions to which the expert system answers using the rules successively until the answer suits the operator.

Alternatively, the assistive device can also use an ontology reasoner. Such an ontology reasoner uses an ontology-type database, defining classes, relationships between classes, and rules. With the application of class relations and rules, the reasoner is able to make inferences.

Alternatively, the assistance device can also use a neural network, the purpose of which is to deduce the function making it possible to link a set of input parameters to different types of output.

However, each artificial intelligence algorithm has limitations that prevent its satisfactory use to solve complex queries involving critical avionics functions.

For example, when the assistance device used is an expert system, if an equivalence relation between reasoning elements is not defined, the expert system is not able to establish the relation between elements. A reasoner on ontology is complex to implement and the computation time is important if the ontology is large. Finally, neural networks can hardly offer an output different from that contained in their learning base, in addition to being difficult to explain and presenting adversarial examples.

There is therefore a need to obtain an electronic decision support device making it possible to propose a solution to the operator's request which is both quick to determine while being relevant and secure for the implementation of the avionics critical function or avionics assistance function.

To this end, the subject of the invention is an electronic decision support device for the implementation of a critical function or an assistance function by an avionic system in response to a request sent by an operator. of the avionic system, said avionic system being capable of operating according to a current context defined by at least one operating parameter of said avionic system, the electronic device comprising:

• a reception module configured to receive the request sent by the operator and to receive the current context,
• a processing module configured to generate at least one recommendation in response to the operator's request, the processing module comprising:

+ a similarity reasoning module configured to generate a recommendation according to the operator's request and the current context from an algorithm based on the content of a reference database, the reference database comprising a list of predetermined contexts, a list of predetermined queries and predetermined recommendations, each recommendation being associated with one of the predetermined contexts and one of the predetermined queries,

+ a rule-based reasoning module configured to generate a recommendation from a deterministic algorithm according to the operator's request and the current context, the deterministic algorithm being made up of a succession of predetermined conditional instructions;

+ an ontology reasoning module configured to generate a recommendation from an ontology-based algorithm based on the operator's query and the current context, the ontology defining a structured set of concepts and relationships between said concepts modeling the operation of the avionics system;

+ an activation module configured to activate at least one of the reasoning modules among the reasoning module by similarity, the reasoning module by rules and the reasoning module on ontology;

a generation module configured to receive at least one recommendation from the processing module and to generate a response to the operator's request from the at least one recommendation received.

According to other advantageous aspects of the invention, the electronic decision support device comprises one or more of the following characteristics, taken in isolation or in all technically possible combinations: - the device further comprises at least one module among:

+ a display module configured to display the response intended for the operator of the avionic system, and

+ a transmission module configured to transmit the response to the avionic system for the implementation of the critical function or the assistance function according to said response;

• each reasoning module is configured to generate an error message sent to the activation module if a recommendation is not generated by said reasoning module after its activation by the activation module, the activation module being configured to successively activating the reasoning module by similarity, the reasoning module by rules and the reasoning module on ontology until one of the reasoning modules generates a recommendation;
• the processing module comprises a preprocessing module configured to perform a semantic analysis of the operator's request according to a predetermined formal rule and activating one of the reasoning modules according to the result of the analysis of the operator's request operator;
• the activation module is configured to activate at least two of said reasoning modules, the processing module further comprising a control module configured to compare the recommendations generated by the at least two reasoning modules and to check whether said recommendations are consistent between they according to a predetermined consistency rule, the generation module being configured to generate the response only if no inconsistency is detected by the control module;
• the similarity reasoning module is configured to compare the query of the operator and the current context with the predetermined queries and the predetermined contexts of the reference database according to a similarity metric, the recommendation generated by the similarity reasoning module being equal to the recommendation associated with the predetermined context and with the predetermined request having the greatest similarity with the current context and the operator's request, the recommendation being generated only if the similarity is greater than a predetermined threshold value; And
• the generation module is further configured to send the generated response associated with the operator's request and the current context to the reference database.

The invention also relates to a control station comprising an electronic device as defined above, the control station being chosen from the group consisting of: a control station arranged in an aircraft; a remote control station for an aircraft, in particular a drone; and a control station disposed in an air traffic ground control station.

The invention also relates to a decision-making aid method for the implementation of a critical function or of an assistance function by an avionic system in response to a request sent by an operator of the avionic system, said system avionics being capable of operating according to a current context defined by at least one operating parameter of said avionics system, the method being implemented by an electronic decision-making aid device as defined above, the method comprising at least the following steps :

• receipt of the request sent by the operator,
• activation of at least one of the reasoning modules from among the reasoning module by similarity, the reasoning module by rules and the reasoning module on ontology;
• generation of at least one recommendation in response to the request,
• receipt of the at least one recommendation and generation of a response to the operator's request from the at least one recommendation received

The invention also relates to a computer program comprising software instructions which, when executed by a computer, implement a method as defined above.

These characteristics and advantages of the invention will appear more clearly on reading the following description, given solely by way of non-limiting example, and made with reference to the appended drawings, in which:

there is a schematic representation of an aircraft and a control station each comprising an electronic device according to the invention,

there is a schematic representation of the electronic device according to the invention,

there is a flowchart of a process, according to the invention, for decision support for the implementation of a critical function by an avionic system implemented by the electronic device of the .

A plurality of avionic electronic systems 10 are represented on the .

The avionic system 10 is in particular embedded in an aircraft 12. The aircraft is typically an airplane, a helicopter, or even a drone. In other words, the aircraft 12 is a flying machine that can be controlled by an operator 14, here a pilot, via a control station 16. The control station 16 is arranged inside the aircraft 12 or else at a distance from the aircraft 12, in particular in the case of a drone.

Alternatively, as also shown in , the avionics system 10 is embedded in a control station 18 on the ground, such as a control tower in an aerodrome for example. The operator 14 controls the air traffic and in particular the aircraft 12 via the control station 16 arranged in the control station 18.

The control station 16 further comprises a man-machine interface 20 configured to receive a request from the operator 14.

The request is in particular a search for information or a search for a solution to a request expressed by the operator 14.

The man-machine interface 20 comprises for example a keyboard, a microphone, etc. The man-machine interface 20 can for example format the request in a predetermined format. As a variant, the request is for example formatted after the reception module 26 introduced later in this description.

The control station 16 further comprises a display screen 22. The display screen 22 is for example a head-down display screen. The screen is then a surface configured to display at least one image. Advantageously, the head-down display screen is configured to display information relating to the aircraft 12, such as the speed, altitude, orientation of the aircraft 12 and/or information relating to the environment exterior of the aircraft 12, such as air traffic information and weather conditions around the aircraft 12.

Alternatively, the display screen 22 is a head-up display screen. The screen is then at least partially transparent. Advantageously, the head-up display screen is a visor integrated into a specific helmet to be worn by the operator 14. Alternatively, the head-up display screen is a transparent surface fixed in the control station 16 and placed in front of the operator 14. As a further variant, the head-up display screen is a windshield of the aircraft 12.

Each avionics system 10 is in particular configured to implement an avionics critical function or an avionics assistance function. The avionics critical function is typically chosen from the group consisting of: a flight control for the aircraft 12, a trajectory calculation for the aircraft 12, a choice of destination aerodrome for the aircraft 12, a regulation order air traffic, a change in communication frequency of the aircraft 12, etc.

In addition or as a variant, each avionics system 10 is configured to implement an avionics assistance function. The avionics assistance function is typically chosen between: a runway highlighting function, an optimized trajectory recommendation for the aircraft 12, an airport recommendation for the aircraft 12 in the event of a flight plan change , etc.

However, those skilled in the art will understand that the invention can also be adapted to other complex requests concerning critical functions or assistance functions, such as vehicle fleet management, a process for controlling nuclear reactor, temperature control in a factory furnace, navigation control of an autonomous motor vehicle, speed control of a railway vehicle, route recommendation for an autonomous vehicle , etc.

Each avionic system 10 is able to operate according to a current context defined by at least one operating parameter associated with said avionic system 10.

Each operating parameter is characteristic data of the current context. The current context corresponds to all of the circumstances in which the avionic system 10 operates.

In particular, each operating parameter associated with the avionics system 10 is chosen from the group consisting of:

- a flight parameter of the aircraft 12 associated with the avionics system 10;

- a flight plan associated with the aircraft 12;

- a mission assigned to the aircraft 12; And

- a meteorological parameter associated with the environment of the aircraft 12,

- a parameter describing the operating state of the avionic system 10.

The flight parameter is in particular the geographical position of the aircraft 12, the altitude of the aircraft 12, the speed of the aircraft 12 or the heading of the aircraft 12.

The flight plan of the aircraft 12 notably comprises a set of waypoints provided for the aircraft 12 between the departure and destination aerodromes.

The mission of the aircraft 12 is the objective of the flight of the aircraft 12, such as transporting passengers and/or goods to a certain destination, reconnaissance or surveillance of an area of interest, etc.

The meteorological parameter is for example the speed of the wind, the presence of bad weather, the visibility of the environment for the pilot, etc.

A parameter describing the operating state of the avionic system 10 is for example a Boolean indicating the self-detection of system failures.

An electronic decision support device 24 for the implementation of a critical function or an assistance function by one of the avionic systems 10 in response to a request sent by the operator 14 of said avionic system 10 is also shown on the .

With reference to the , the electronic device comprises a reception module 26, a processing module 28 and a generation module 30.

Advantageously, the electronic device 24 further comprises a display module 32, a transmission module 34.

The reception module 26 is configured to receive the request transmitted by the operator 14. In particular, the reception module 26 is configured to receive the request formatted by the man-machine interface 20 and to transmit it to the processing module 28.

The reception module 26 is further configured to receive the current context. In particular, the reception module 26 is configured to receive the operating parameter(s) associated with the avionic system 10 and making it possible to characterize the current context. The reception module 26 is suitable for receiving, for example, the flight parameters by the flight control system or the flight plan by the FMS (“Flight Management System” in English or Flight Management System in French).

The processing module 28 is configured to generate at least one recommendation in response to the request from the operator 14.

The recommendation is in particular a suggested action on one of the avionic systems 10, in particular a command acting on the avionic system 10 in order to modify its operation in response to the request of the operator 14.

The operator's request 14 for example follows an event potentially affecting the safety of the aircraft 12. The event is for example an event external to the avionics system 10 presenting a possible risk for the safety of the avionics system 10 or else an event internal to the system 10, such as the failure of one of the components of the avionics system 10.

The processing module 28 includes three reasoning modules and an activation module 36.

Each reasoning module is configured to generate a recommendation according to the request from the operator 14 and the current context received by the reception module 26.

The three reasoning modules are in particular a reasoning module by similarity 38, a reasoning module by rules 40 and a reasoning module on ontology 42.

The reasoning module by similarity 38 is configured to generate a recommendation according to the request of the operator 14 and the current context from an algorithm based on the content of a reference database 44.

The reference database 44 includes a list of predetermined contexts, a list of predetermined queries and predetermined recommendations. Each recommendation is associated with one of the predetermined contexts and one of the predetermined queries.

Thus, the reference database 44 comprises a set of requests and contexts already encountered or already considered upstream of the operation of the avionic system(s) 10 as well as the associated recommendation. The content of the reference database 44 is established, prior to the operation of the avionic system 10, so that each recommendation complies with the avionic safety rules applying to the avionic system 10.

The reasoning module by similarity 38 is configured to compare the query of the operator 14 and the current context to the predetermined queries and to the predetermined contexts contained in the reference database 44 according to a similarity metric.

The similarity metric is a function characterizing the similarity between two contexts and between two queries, in particular by comparing the different operating parameters of the two contexts and by comparing the keywords of the query.

The similarity metric makes it possible to associate with each pair {current context, query of the operator} a measure of similarity with the various pairs {context, query} contained in the reference database 44. Plus the value of the measure is high the more the two pairs are similar according to the similarity metric used.

The similarity metric is for example the so-called cosine similarity function. Cosine similarity is used to calculate the similarity between two vectors to be compared. The cosine similarity is equal to the dot product of the two vectors, divided by the norm of the two vectors. The result is therefore between -1 and 1. The value -1 indicates that the vectors are opposite, the value 0 indicates that the vectors are independent and the value 1 indicates that the vectors are similar, in particular collinear. The intermediate values between -1 and 1 make it possible to evaluate the degree of similarity between the two vectors.

Thus, the reasoning module by similarity 38 is configured to calculate the value of the similarity metric between the pair {current context, query of the operator} and the different pairs {context, query} contained in the reference database 44, then to sort the {context, query} pairs from the most similar to the least similar with respect to the {current context, operator query} pair from the calculated metric values.

The recommendation generated by the reasoning module by similarity 38 is then equal to the recommendation associated with the predetermined context and with the predetermined request having the greatest similarity with the current context and the request of the operator 14. In other words, the reasoning by similarity 38 selects the pair {context, query} contained in the reference database 44 whose measure of similarity with the pair {current context, query of the operator} is the highest.

Nevertheless, the generation of the recommendation by the similarity reasoning module 38 is inhibited if the similarity value is less than a predetermined threshold value. In other words, the recommendation is generated only if the similarity is greater than the predetermined threshold value.

The predetermined threshold value is determined so as to limit the security risks for the avionic system 10 and makes it possible to ensure that the recommendation generated is sufficiently relevant to respond to the request of the operator 14 in view of the current context.

The reasoning module by similarity 38 is configured to generate an error message sent to the activation module 36 if a recommendation is not generated, in particular if the greatest similarity obtained is lower than the predetermined threshold value. Indeed, this means that the reference database 44 does not contain a pair {context, query} sufficiently similar to the current situation.

By way of example, in the context of a request aimed at determining the landing runway of the aircraft 12, the reference database 44 notably comprises the following data represented in the table below.

Query Context Solution No. Track requested Runway n°1 open Runway n°2 open Runway n°3 open Chosen track 1 1 1 1 1 1 2 1 0 1 1 2 3 2 0 0 1 3 4 2 1 0 1 3

When the aircraft 12 is in the current context "requested runway=1 and runway n°1 open=0 and runway n°2 open=1" then the reasoning module by similarity 38 determines that the context of line 2 of the tableau is the most similar to the current context and thus generates a recommendation to choose track #2.

In particular, in the present case, the similarity metric is a measure of the number of columns corresponding exactly to the current context. Line 1 here corresponds to a similarity value of 2, line 2 corresponds to a similarity value of 3, line 3 corresponds to a similarity value of 1 and line 4 corresponds to a similarity value of 0. So , the highest similarity value is that corresponding to line 2 and therefore the recommendation chosen is that associated with this line.

Here, the value of the predetermined threshold is to be at least equal to 3 and thus the reasoning module by similarity 38 indeed generates the recommendation to choose track n°2.

The reasoning module by rules 40 is configured to generate a recommendation from a deterministic algorithm according to the request of the operator 14 and the current context.

The deterministic algorithm consists of a succession of predetermined conditional instructions.

In particular, the deterministic algorithm consists, for example, of a succession of “if, then” type rules. These rules are in particular constructed by a person skilled in the art or deduced from data resulting from experience feedback from the avionic system(s) 10, for example data contained in the reference database 44.

If the conditional instructions lead to a solution, the latter is then generated as a recommendation by the rule-based reasoning module 40.

If the conditional instructions lead to an empty solution set then the rule-based reasoning module 40 is configured to generate an error message sent to the activation module 36 indicating that a recommendation is not generated

By way of example, for the same case of a request aimed at determining the landing runway of the aircraft 12, the predetermined conditional instructions are:

1. Rule: Calculation of the runway distance necessary for landing according to the dynamics of the aircraft 12, in particular its speed and altitude;
2. Rule: If the runway distance i is greater than the required runway distance then the landing is said to be “OK”;
3. Rule: If several runways i allow a so-called “OK” landing, choose the runway with the highest runway distance.

In this context, the rule-based reasoning module 40 successively applies rules 1, 2 then 3 to try to determine a solution to the request. For example, if the application of rule 1 leads to a necessary runway distance of 100 m and three runways are nearby, including a runway A with a length of 150 m and a runway B with a length of 120 m and a runway C of length 80 m, then the reasoning module by rules 40 excludes track C by application of rule 2 then proposes track A by application of rule 3.

The ontology reasoning module 42 is configured to generate a recommendation from an ontology-based algorithm according to the query of the operator 14 and the current context.

The ontology defines a structured set of concepts and relations between said concepts modeling the operation of the avionic system 10. In particular, the ontology is a semantic network which groups together a set of concepts linked to each other by taxonomic relations in order to prioritize concepts on the one hand, and semantics on the other. Thus, knowledge is structured in the form of a model formalized in descriptive logic. A reasoner on ontology makes it possible to deduce by inference more than the knowledge strictly written in the initial ontology.

The ontology reasoning module 42 is configured to generate an error message sent to the activation module 36 if a recommendation is not generated, in particular if the ontology does not make it possible to deduce a recommendation.

By way of example, still within the framework of a request aimed at determining the landing runway of the aircraft 12, the ontology comprises an “airport” class with “runways” subclasses. The airport class can for example have name properties. The track subclass contains in properties the name, the length of the track and its state, namely closed or open. There are X "airport" instances relating to the X modeled airports. There are Y “runway” instances relating to airport runways. The "airport" instances and the "runway" instances are linked by a property {airport instance} a_for_runway {runway_instance}. The ontology also includes an “aircraft” class with the properties of aircraft name, aircraft speed, minimum runway distance to allow landing, the name of the planned landing runway. The current aircraft 12 is represented in this ontology by an instance, with the properties fulfilled, in particular by means of the current context. The "aircraft" instance is linked to a runway instance by the "going_to_land" property. Following the transmission of the request to “find runway for landing”, the ontology reasoning module 42 transmits a request to ontology asking it to look for the “runway” instance such as:

• runway length [runway instance] < minimum distance [aircraft instance], and
• track status of the “track” instance = Open.

The activation module 36 is configured to activate at least one of the reasoning modules among the reasoning module by similarity 38, the reasoning module by rules 40 and the reasoning module on ontology 42.

Following its activation by the activation module 36, said reasoning module generates a recommendation or generates an error message if a recommendation is not likely to be generated.

According to an advantageous embodiment of the invention, the activation module 36 is configured to successively activate the reasoning module by similarity 38, the reasoning module by rules 40 and the reasoning module on ontology 42 until one of the reasoning modules generates a recommendation.

Thus, following the reception of a request from the operator 14, the activation module 36 activates the reasoning module by similarity 38. If the reasoning module by similarity 38 generates a recommendation without generating an error message, the processing module 28 transfers this recommendation to the generation module 30. Conversely, if the reasoning module by similarity 38 generates an error message, the activation module 36 then activates the reasoning module by rules 40. Similarly, if the reasoning module by rules 40 generates a recommendation, this is sent to the generation module 30 and if an error message is generated, the activation module 36 finally activates the reasoning module on ontology 42 .

The activation module 36 therefore makes it possible to adapt the reasoning load to the complexity of the request of the operator 14. The more complex the request, the more the complexity of the reasoning module making it possible to solve the request is necessary. Indeed, a recommendation is quickly obtained by the reasoning module by similarity 38 when the situation is already known, that is to say that a query and a similar current context are present in the reference database. 44 and that a predetermined recommendation is associated with it. In this case, no calculation is necessary, in addition to the calculation of the similarity which does not require complex calculations. When the situation does not exist in the reference database 44, reasoning based on simple rules is carried out in an attempt to obtain a recommendation. Finally, when the situation is not encountered and the rule-based reasoning module 40 cannot provide a solution, then the ontology-based reasoning is carried out to find a solution to the query and obtain a recommendation.

As a variant or in addition, the processing module 28 further comprises a pre-processing module 50 configured to perform a semantic analysis of the request from the operator 14 according to a predetermined formal rule when the request is received by the processing module. 28 and activate one of the reasoning modules according to the result of the analysis of the request of the operator 14.

In particular, the pre-processing module 50 is configured to analyze the keywords of the request from the operator 14 and thus determine if the request has a sufficient probability of already being known to the reference database 44, or if this is not the case, the request has a sufficient probability of being correctly processed by the rule-based reasoning module 40. For example, the formal rule is associated with a predetermined list of keywords each associated with one reasoning modules.

Thus, following receipt of a request from the operator 14, the preprocessing module 50 analyzes this request and determines, in particular by means of the keywords of the request, that the reasoning module by similarity 38 has little chance of generating a recommendation and directly activates the rule-based reasoning module 40. Then, if the rule-based reasoning module 40 generates a recommendation, it is sent to the generation module 30 or if an error message is generated, the module activation 36 activates the ontology reasoning module 42.

As a variant or in addition, the activation module 36 is configured to activate at least two of said reasoning modules.

In particular, the activation module 36 is configured to activate one of the reasoning modules and the reasoning module of just greater complexity. For example, if the activation module 36 activates the reasoning by similarity module 38, it also activates the reasoning by rules module 40. If the activation module 36 activates the reasoning by rules module 40, it also activates the reasoning module on ontology 42.

The processing module 28 then further comprises a control module 52 configured to receive the recommendations generated by the two reasoning modules and to compare them with each other. The control module 52 is configured in particular to check whether said recommendations are mutually consistent according to a predetermined consistency rule. The consistency rule makes it possible to determine whether the two recommendations generated aim at a similar, or even identical, implementation of the critical function or the assistance function.

In the example of a query to determine the landing runway for aircraft 12, two recommendations are consistent if they recommend the same landing runway.

The control module 52 is configured to send the generation module 30 an inconsistency message if an inconsistency is detected.

The generation module 30 is configured to receive the recommendation from the processing module 28 and to generate a response to the request from the operator 14 from the recommendation received.

The response is in particular a solution to the request for information from the operator 14.

As a variant or in addition, the response is an instruction to be implemented by the avionic system 10 in response to the request from the operator 14.

In particular, the response corresponds to the recommendation received, formatted to be possibly applied by the corresponding avionic system 10 .

The generation module 30 is configured to generate the response only if no inconsistency is detected by the control module 52.

Advantageously, the generation module 30 is also configured to send the generated response associated with the request of the operator 14 and the current context to the reference database 44, in particular when the associated recommendation is generated by the reasoning module by rules 40 or by the reasoning module on ontology 42. Thus, the reference database 44 comprises an additional situation and the reasoning module by similarity 38 is likely to the next similar request in a similar current context to directly determine the associated recommendation.

The generation module 30 is configured to send the generated response to the display module 32 or to the transmission module 34.

The display module 32 is configured to display the response intended for the operator 14, in particular on the display screen 22. In particular, the display module 32 is configured to display the response on the screen. head-down display or the head-up display screen in front of the operator 14 who possibly applies the instruction associated with the response.

By way of example, the display module 32 displays the number of the landing runway determined by the processing module 30.

As a variant or in addition, the display module 32 is configured to display a button allowing the user to accept or not the instruction associated with the response. The electronic device then further comprises an acquisition module configured to acquire the choice of the operator and, in the event of acceptance by the operator of the instruction, to send the instruction to the transmission module 34.

Thus, if the operator 14, here the pilot, accepts the proposed landing strip, the response is sent to the FMS.

The transmission module 34 is configured to transmit the response to the corresponding avionic system 10 for the implementation of the critical function or the assistance function according to said instruction.

Thus, the transmission module 34 is configured to directly send the response as soon as it is generated by the generation module 30 to the avionics system 10 for the implementation of the critical function or the assistance function in accordance with the request of the operator 14.

In the example of the , the electronic device 24 comprises an information processing unit formed for example of a memory and a processor associated with the memory. The reception module 26, the processing module 28, the generation module 30 and advantageously, the display module 32 and the transmission module 34 are each produced in the form of software, or a software brick, executable by the processor. The memory is then suitable for storing reception software, processing software, generation software, and in optional addition, display software and transmission software. The processor is then capable of executing each of these software programs.

In a variant not shown, the reception module 26, the processing module 28, the generation module 30 and advantageously, the display module 32 and the transmission module 34, are each made in the form of a programmable logic component, such as an FPGA (Field Programmable Gate Array), or else in the form of a dedicated integrated circuit, such as an ASIC (Application Specific Integrated Circuit).

When the electronic device 24 is produced in the form of one or more software, that is to say in the form of a computer program, it is also capable of being recorded on a medium, not shown, readable by computer. The computer-readable medium is, for example, a medium capable of storing electronic instructions and of being coupled to a bus of a computer system. By way of example, the readable medium is an optical disc, a magneto-optical disc, a ROM memory, a RAM memory, any type of non-volatile memory (for example EPROM, EEPROM, FLASH, NVRAM), a magnetic card or an optical card. On the readable medium is then stored a computer program comprising software instructions.

The operation of the electronic decision aid device 24 according to the invention will now be explained using the representing a flowchart of the method, according to the invention, of decision support for the implementation of a critical function or of the assistance function by an avionic system 10 in response to a request sent by an operator 14 of the avionics system 10.

Initially, an operator 14 is installed in the control station 16.

The control station 16 is installed in an aircraft 12 or on the ground, for example in a control station 18, as represented on the .

During an initial step 100, the reception module 26 receives a request sent by the operator 14. The request is for example a request for allocation of a landing strip for the aircraft 12. As another example , the request is a request to modify the flight path of the aircraft 12. As yet another example, the request is a request for a reconfiguration solution following the detection of a failure of an avionics system 10.

During an optional step 110, the preprocessing module 50 analyzes the request, according to a formal rule, in particular by searching for keywords.

Then, during a step 120, the activation module 26 activates at least one of the reasoning modules, in particular the reasoning module corresponding to the analysis carried out by the preprocessing module 50.

During a step 130, a recommendation is generated in response to the request.

As a variant, the method does not include step 110 and during step 120, the activation module 36 is configured to successively activate the reasoning module by similarity 38, the reasoning module by rules 40 and the reasoning on ontology 42 until one of the reasoning modules 38, 40, 42 generates a recommendation during step 130. This recommendation is then sent to the generation module 30.

Thus, the activation module 36 activates the reasoning module by similarity 38 during a sub-step 121. If the reasoning module by similarity 38 generates a recommendation without generating an error message during a sub-step 131 , the processing module 28 transfers this recommendation to the generation module 30.

Conversely, if the reasoning module by similarity 38 generates an error message, the activation module 36 then activates the reasoning module by rules 40 during a sub-step 122.

Similarly, if the rule-based reasoning module 40 generates a recommendation during a sub-step 132, the latter is sent to the generation module 30.

If an error message is generated by the reasoning module by rules 40, the activation module 36 finally activates the reasoning module on ontology 42 during a sub-step 123 and the reasoning module on ontology 42 then generates a recommendation during a sub-step 133.

Then, during a step 140, the generation module 30 receives the recommendation and generates a response to be implemented by the corresponding avionics system 10 from the recommendation received.

During a step 150, the response is sent to the display module 32 which displays the response in front of the operator 14 on the display screen 22.

As a variant or in addition, the response is sent to the transmission module 34 which transmits the setpoint associated with the response to the avionic system 10 for the implementation of the critical function or the assistance function according to the setpoint, when a step 160.

It can then be seen that the present invention has a certain number of advantages.

Indeed, the invention allows the generation of a relevant answer concerning the implementation of a critical function or an assistance function by adapting the complexity of the algorithm used to the complexity of the request of the operator 14, in particular by means of the activation module 36 making it possible to adapt the reasoning load to the complexity of the request of the operator 14.

When the situation is already known, a recommendation is obtained very quickly by the reasoning module by similarity 38. When the situation is not known, the reasoning module by rules 40 is used to deduce the solution to be proposed. Finally, when the situation is not encountered and the application of the rules cannot provide a solution, then the reasoning on the basis of ontology is carried out in order to obtain a recommendation.

Thus, a parallel can be drawn between the electronic device according to the invention and the cognitive model of human reasoning described by Rasmussen. This cognitive model is called “SRK” from English “Skills, Rules, Knowledge”, or “Habiletés, Rules, Savoirs” in French. It describes reasoning in the form of three elements of increasing complexity: the automatic skills achieved in the context of simple tasks, the rules of calculation resulting from routines and the knowledge resulting from complex reasoning, used in the event of confrontation with the unknown.

The electronic decision support device 24 according to the invention makes it possible to propose a solution to the request of the operator 14 which is relevant, while being compatible with the safety of the associated avionic system 10.

Claims (10)

Electronic decision support device (24) for the implementation of a critical function or an assistance function by an avionics system (10) in response to a request transmitted by an operator (14) of the system avionics system (10), said avionics system (10) being able to operate according to a current context defined by at least one operating parameter of said avionics system (10),
the electronic device (24) comprising:

• a reception module (26) configured to receive the request sent by the operator (14) and to receive the current context,
• a processing module (28) configured to generate at least one recommendation in response to the request from the operator (14), the processing module (28) comprising:

+ a similarity reasoning module (38) configured to generate a recommendation according to the operator's request (14) and the current context from an algorithm based on the content of a reference database ( 44), the reference database (44) comprising a list of predetermined contexts, a list of predetermined queries and predetermined recommendations, each recommendation being associated with one of the predetermined contexts and one of the predetermined queries,
+ a rule-based reasoning module (40) configured to generate a recommendation from a deterministic algorithm according to the operator's request (14) and the current context, the deterministic algorithm being made up of a succession of pre-determined conditional statements;
+ an ontology reasoning module (42) configured to generate a recommendation from an ontology-based algorithm based on the query of the operator (14) and the current context, the ontology defining a structured set of concepts and relationships between said concepts modeling the operation of the avionics system (10);
+ an activation module (36) configured to activate at least one of the reasoning modules among the reasoning module by similarity (38), the reasoning module by rules (40) and the reasoning module on ontology (42 );

• a generation module (30) configured to receive at least one recommendation from the processing module (28) and to generate a response to the request from the operator (14) from the at least one recommendation received.

Electronic device (24) according to claim 1, further comprising at least one of:
- a display module (32) configured to display the response intended for the operator (14) of the avionic system (10), and
- a transmission module (34) configured to transmit the response to the avionic system (10) for the implementation of the critical function or the assistance function according to said response. An electronic device (24) according to claim 1 or 2, wherein each reasoning module (38, 40, 42) is configured to generate an error message sent to the activation module (36) if a recommendation is not generated by said reasoning module (38, 40, 42) after its activation by the activation module (36),
the activation module (36) being configured to successively activate the reasoning by similarity module (38), the reasoning by rules module (40) and the reasoning on ontology module (42) until one reasoning modules (38, 40, 42) generates a recommendation. An electronic device (24) according to claim 1 or 2, wherein the processing module (28) comprises a pre-processing module (50) configured to perform a semantic analysis of the operator's request (14) according to a predetermined formal rule and activating one of the reasoning modules (38, 40, 42) according to the result of the analysis of the request from the operator (14). An electronic device (24) according to claim 1 or 2, wherein the activation module (36) is configured to activate at least two of said reasoning modules (38, 40, 42), the processing module (28) comprising in addition to a control module (52) configured to compare the recommendations generated by the at least two reasoning modules (38, 40, 42) and to check whether said recommendations are consistent with each other according to a predetermined consistency rule,
the generation module (30) being configured to generate the response only if no inconsistency is detected by the control module (52). An electronic device (24) according to any preceding claim, wherein the similarity reasoning module (38) is configured to compare the query of the operator (14) and the current context to the predetermined queries and the predetermined contexts of the reference database (44) according to a similarity metric,
the recommendation generated by the reasoning module by similarity (38) being equal to the recommendation associated with the predetermined context and with the predetermined request having the greatest similarity with the current context and the request of the operator (14),
the recommendation being generated only if the similarity is greater than a predetermined threshold value. Electronic device (24) according to any one of the preceding claims, in which the generation module (30) is further configured to send the generated response associated with the request from the operator (14) and the current context to the base reference data (44). Control station (16) comprising an electronic device (24) according to any one of the preceding claims, the control station (16) being chosen from the group consisting of:

• a control station (16) disposed in an aircraft (12),
• a remote control station (16) of an aircraft (12), in particular of a drone,
• a control station (16) disposed in an air traffic ground control station (18).

Decision-making aid method for the implementation of a critical function or an assistance function by an avionic system (10) in response to a request sent by an operator (14) of the avionic system (10) , said avionic system (10) being capable of operating according to a current context defined by at least one operating parameter of said avionic system,
the method being implemented by an electronic decision aid device (24) according to any one of claims 1 to 7,
the method comprising at least the following steps:

• receipt (100) of the request sent by the operator (14),
• activation (120) of at least one of the reasoning modules among the reasoning module by similarity (38), the reasoning module by rules (40) and the reasoning module on ontology (42);
• generation (130) of at least one recommendation in response to the request,
• receiving the at least one recommendation and generating (150) a response to the request from the operator (14) based on the at least one recommendation received.

Computer program comprising software instructions which, when executed by a computer, implement a method according to the preceding claim.