FR2994678A1 - Method for assisting train driver for driving train, involves performing task analysis to evaluate and transmit summary of task, and determining curvilinear abscissa of train head, where curvilinear abscissa is transmitted with task - Google Patents

Abstract

The method involves performing a task extraction sequence (3x), where the task extraction sequence transmits information from sources (1x). A task analysis is performed (4) to evaluate and transmit a summary of a task. A curvilinear abscissa of a train head is determined (6), and the curvilinear abscissa is transmitted with the task. A train signal is determined (7), and a warning is issued (9) for an imminent braking. A braking assistance is performed (11), and an optimal speed of the train is determined (12).

Description

The present invention relates to a method that uses in real time the information provided by several geolocation sources to evaluate the reliability of systems internal or external to the process, to determine the position and speed of a train on the track, calculate a set of elements useful to the successful completion of the mission, the process adapting to the drivers and meeting the requirements of the profession of train driver. [1] The use of embedded geolocation systems has already been the subject of numerous patent applications in the field of railway safety, for example: US Pat. No. US6081769 to Dwight D. CURTIS who proposes to determine the length of a train using a GPS receiver at each end of the train, the assembly being coupled to a calculator, patent No. US6459965 of Alan L. POLIVKA who proposes to locate the positions of several rail convoys to prevent risks of collisions.

A constant concerning the majority of patents filed is that they use a plurality of GPS receivers embedded on a train, considering that each receiver gives usable information whose error is controllable. The very nature of a train however imposes the use of GPS receivers close to the ground so close to obstacles that alter in many ways the quality of the signal, obstacles to this signal that succeed each other at the speed of the train. A rail GPS tool must therefore take into account this fact, or there will be a random component that will lead the driver of a train to work with a tool with questionable reliability. [2] Mark KANE and Harrison Thomas HICKENLOOPER's EP2178734 (A1) patent represents the closest technical state of the present invention. 1) It proposes for each satellite to quantify certain errors due to atmospheric degradation by using two known ground-based GPS receivers with known fixed positions, then to transmit this information to an onboard subsystem consisting of a computer and two spaced GPS receivers. from a known distance. Various methods for detecting other errors are then applied: a calculation of heading carried out as a function of the positions of the two onboard receivers, an error estimate being made by comparing this heading with a database containing vectors tangent to the track in a multitude of 35 points, a calculation of the GPS distance between the two onboard receivers, an error estimate being made by comparing it with the known actual distance, a comparison of the speeds detected at the same time by the two onboard GPS units. , the use of beacons implanted along the track, tags which record the date of the passage of the train and transmit this date to the calculator which retrieves the GPS position information of the train recorded on the same date and compares it with the position of the tag crossed, a comparison of GPS speed information compared to that of a speed calculator connected to the wheels.

In order for the error due to the atmospheric degradations calculated at the fixed stations level to be comparable to that concerning the receivers situated in the train, such a method requires a proximity between the train and the fixed stations. 2) To position the train on the track, patent EP2178734 (A1) proposes to use a database containing staking points along the track and to determine a first point corresponding to the nearest point that the train comes to pass on its course, and a second point corresponding to the nearest point that the train has not yet passed on its course, then to calculate the distance between the GPS position selected for the train and the line passing through said two points .

However, traffic close to major cities, such as the one around Paris, requires the use of kilometers of roads with curves often pronounced which could generate significant lateral errors. Patent EP2178734 (A1) proposes as a solution to increase the density of points present in the database, multiplying the number of points in the same curve. However, for some curves, if this method generates a significant lateral error, compensating for it correctly requires entering into the database points spaced a distance of less than twice the average accuracy error that this DGPS method can provide without any attempt. resetting on the track. This could make it inappropriate if the system is assigned the mission of providing a driver with speed limit information between two specific points on the track, a braking signal or other information that requires good positioning accuracy. the way and deadlines. 3) With the method described in patent EP2178734 (A1) these positioning errors on the track are amplified when the real distance between the two points retained in the database tends towards the distance error that exists between the coordinates GPS retained and the actual coordinates. - 3 - 4) This error in determining the position of the train on the track is further magnified if the GPS positioning error causes the system to consider a point as crossed when it is not. To this, the patent EP2178734 (A1) proposes as a solution the implementation of beacons on the ground to the right of the points recorded in the database, beacons that locate the passage of the train and validate the GPS information. It also proposes the use of an odometric validation system. However, these two more precise methods are expensive and could disqualify the interest of using a GPS system in nominal mode. [003] In order to use GPS information in the field of train driving, it is necessary to attach great importance to: the information determination method, the information validation method, the train position determination method on the way from the selected information, to the costs of implementing said methods and to the interest they arouse with respect to an adaptation of the existing systems, to the proposed functionalities when these methods are intended to serve a driver of trains and its hierarchy in their missions. [004] Experience feedback concerning railway incidents of types: accidental crossing of stop signs, omissions or bad executions of commercial stops, authorized speed overruns, non-compliance with specific safety rules, shows that unsuitable behavior of the driver to a situation is due in a very large majority of cases to: a sequence break, and / or a poor representation of the position of the railway train on the line with respect to a safety device (signal d 'stop, speed limit zone or area of application of a point of the Regulation), and / or a poor representation of train characteristics: speed, composition, braking performance. [005] The term "sequence break" is derived from the computer language. In the field of railway driving, it refers to the intrusion of one or more events in the course of a process carried out by the driver of a train and, depending on the complexity of the event and the cognitive resources implemented to manage it, the mask action it generates on the elements of the process. This can lead to forgetting the contexts that led to the implementation of said method. [6] The break in sequence may lead to forgetting to make a commercial stop, forgetting the imminence of the meeting with a stop signal, and more generally forgetting to apply a rule prescribed by regulatory documents related to traffic safety.

The driver's skills and analytical abilities are never blamed when a sequence break occurs and its consequences, so all drivers can experience it unhappily. This is why upstream a number of business processes called "hands-on" are implemented by drivers. Some drivers even create "sticky notes", such as cards on which are written information, cards placed on the desk if necessary. But a file may inadvertently be covered by another document, it can be forgotten at home, and not see it can be perceived as the absence of danger despite the situation. In the same vein, efficiency is only guaranteed when it is fully implemented. [7] Concerning the accidental crossing of stop signals, technical devices have been implemented to act on the train downstream of a sequence break to allow to avoid the consequences in terms of railway accidents. Their actions are carried out for some of them by acting: directly on the emergency braking of the train, such as the KVB system, on the emergency braking via the driver, such as the detonating cartridge of a square signal "closed ", Which explodes when a train crosses said signal, the driver then engaging the emergency braking system.

These are often the last resort. Thus, the embedded system KVB has been designed to avoid the commitment of a specific point and not to prevent the crossing of the upstream signal whose "closure" however requires the driver to stop before. The KVB will beep alert if the speed of the convoy exceeds that of an alert curve positioned upstream of the point to be protected and will stop the emergency train if said speed exceeds an emergency braking curve. However, the powers developed by the modern accelerating equipment and the reaction times of the brake equipment are such that a reflex action on the brake control by the driver following the emission of the beeps of KVB alerts will only very exceptionally prevent the crossing of the emergency braking curve. We can see the reason in the LOPI report published in 2008 and dealing with KVB: "(The KVB) is not" security ": development time considerations, economic and technical constraints have made that the KVB n was not designed with the features of a security system. must not be in series in the brake signal chain and remain transparent for drivers as long as they comply with the rules of conduct. It automatically triggers emergency braking if it does not. [008] Note then that a number of closed signal crossings protecting dangerous points follow an approach phase of said signals less than 30km / h, a threshold imposed by the operation of KVB this 200 meters upstream of the signal and up to signal This is a speed below which the KVB will not intervene (or 10Km / h when the distance is reduced between the signal and the point to be protected). This speed imposed by the KVB is called Safe Approach Speed or VISA. In a large majority of cases, a driver who, under these conditions, passes a closed stop signal has been the victim of a combination of factors linking: sequence break, loss of attention, poor estimation of the position of the train relative to at the signal, this after he obtained a speed of said train less than 30km / h. [009] Note then that any inadvertent crossing of a closed stop signal is subject to heavy conservatory measures for the purposes of an investigation, measures which have financial and human consequences for the company, its staff, and its customers. [010] Note that given their implementation costs some systems, such as the KVB, do not equip the entire network. [011] An emergency braking has a cost in terms of: - wear of the brake shoes, - energy to revive the train speed, - time lost by this train, - 6 - impact on the other traffic behind this train or waiting on other routes, this impact is reflected in terms of energy and time lost, immobilization of the train and other trains impacted, to change their brake components degraded prematurely, d image of the customer. [12] The time that elapses between the break of sequence and the moment when the train crosses the curve which imposes the emergency braking can last several tens of seconds or even several minutes according to the situations. A delay so far no process 10 exploits in a sustainable manner to cause the driver to perceive his error. [13] Experience feedback also shows that each break in the sequence that results in a missed commercial stop occurs either during the previous stop or during the trip between the two stops. However, the current embedded systems are not intended to remind the driver of the position of the train on the course or the next commercial stop to be made, except the SNCF SIRIUS tool which puts this information in a mode of use called "mode conduct ". It should be noted that the work of mental representation of the next commercial stop is done essentially during the previous stop, and later if the driver feels a doubt. Thus the break in sequence that occurs between the moment when the driver has mentally figured out his next stop (via SIRIUS or any other means) and the moment when he has to brake to make this stop, will probably result in a failure to 'stop. [14] Stop before stop signal and stoppage are safety-specific rules. There is also the Marche à Vue and the Prudente Walk. [15] The SOW requires the driver to adjust the speed of his train so that he can stop before a stop signal, a train tail or an obstacle, and not to exceed the speed of 30km / h. This rule applies in a number of 30 cases, including that of stopping in full lane, ie off-shore planned to be served, stop not motivated by a stop signal, this stop being greater than 3 minutes. Even if this is not prescribed, some drivers begin to apply the Rule of View 200 meters upstream of a "closed" signal, this self-conditioning allowing them to psychologically mark the fact that they must lend. attention to the next signal or they must stop before if necessary. It is also a business adaptation to avoid forgetting the rule of VISA imposed by the KVB at the approach of the signal. But, as we saw in [006], a break in Sequence or loosening of attention may lead unexpectedly to a closed stop signal. In addition, a stop on the way out of any signal is often due to security or technical requirements that can lead the driver to apply various documents and solve more or less complex problems, with sometimes an external pressure related to travelers or the regularity of the traffic. As the feedback from experience shows, finding the solution to a technical or human problem can cause the driver to forget to apply the rule of the COW while this rule allows, for example, to mitigate the consequences of cases where the switches would have been modified to allow other trains to run. Some drivers have therefore created "sticky notes" such as a small card on which is inscribed the information "Step by Step" or "MAV", cardboard placed on the desk if necessary. But a cardboard can inadvertently be covered by another document, it can also be forgotten at home, and not see it can be perceived as the absence of a Stroke to apply despite the situation. [16] The Prudent Walk requires the driver to adapt the speed of his train to a portion of the route according to a known pattern (for example, a presumption of obstacle on the track). The order is transmitted by means of a paper document or by radio or telephone, etc. Some means of transmission therefore require the driver to retranscribe the order and collationner. Experience feedback shows that errors can be generated: at the time of transcription of the order and its collation, by breaking the sequence on the path that leads the train to the area defined by the order, by obscuring the document by another document, by misregistration of said area due for example to milestones and hectometres in poor condition. Finally, it should be noted that once on the zone, the search for a danger can be overshadowed by the natural desire to search for an element of infrastructure materializing the end of the zone. To date there is no method to overcome these difficulties. [17] The representation of a driver of his train on the tracks is defined by the position of the head and the position of the tail of the convoy on the tracks, in relation to: 35 the infrastructures, the zones defining speeds not to not to exceed - 8 - - to the zones imposing the application of specific points of the regulation, A bad spatial representation of the train will be able among other things to have a stop at a bad place on a quay, The feedback of experiment shows that an erroneous representation of the composition of the train is made either very early during the driving of said train, for example from the first turns of wheels which make it leave the station of departure, even before this departure, or later when the driver confuses his train with another train he drove during the same day of service. Commuter train drivers are used to referring to technical tips to remember the composition of their trains but few of them are effective, some may even mislead them or even lead them to generate incidents. technical orders. For example: the opening of the circuit breakers which causes the display of a pictogram indicating the number of engine blocks not powered, display which coincides on some trains with the number of cars making up the train. The SNCF "SIRIUS" tool takes the length of the train on its screen in driving mode, its number of vehicles, which is an effective solution to answer a doubt occurring during a trip, but inefficient if the error occurs upstream or that the driver is persuaded not to go wrong.

Note that on line C of the RER some drivers have small boxes "4V", "6V", "8V" they put on the desk to be able to remember at a glance the composition of their trains. But such cartons can inadvertently be covered by other documents, they can also be forgotten at home. [018] Note that an incident can occur on a construction site area (work zone on the track) if the driver has an inaccurate representation of his train. This representation may be due to kilometric and hectometric markers in bad or missing states that prevent drivers from properly locating themselves.

In such situations, an emergency stop of the KVB is a late action that aims to overcome the consequences of such inaccurate representations, but in no way to prevent the causes while upstream one could help the driver who is in the error in remembering the composition of his train and its position on the track which would avoid an expensive action of the KVB, the SACEM system that equips some trains of the RATP allows to inform and - 9 - to remind the driver the speed not to be exceeded on the track section, it is an expensive system, the TVM, which equips the trains at high speed, serves as on-board signaling and makes it possible to indicate to the driver the speed not to be exceeded, he This is also an expensive system, some engines are equipped with a system called Speed Imposée which allows to set a speed stop not to exceed. But this cleat is a function of a good mental representation of the train on the network by the driver and therefore prone to human error, moreover it happens that the system of Imposed Speed works poorly. [019] A poor perception of the train entity in terms of speed, composition, braking performance may lead the driver to adopt an unsuitable behavior to a situation: braking and acceleration late or too early, too pronounced, which may result in: out-of-wharf stops, braking curve overruns with KVB support, broken couplings. [020] Providing a driver with information that causes him to modify the state of the environment by his action on commands requires that the information transmitted be correct in terms of value and exploitability over time. [21] The information transmitted by a satellite to a GPS receiver is naturally and artificially degraded. When a system has the purpose of transmitting security information to an operator, one must make an effort to master the foreseeable error and to contain the effects due to unforeseeable errors. [22] A GPS receiver is a device that can fail. In such a case it is preferable to have a redundant system capable of detecting a failure and to remedy it where possible. [23] The present invention aims to overcome these disadvantages, and proposes a method. [024] The accompanying drawings illustrate the invention: FIG. 1 is a representation of the hierarchy of information transfers between the tasks of the invention, with respect to time. FIG. 2 is a representation of the hierarchy within the invention of all information transfers, with respect to time. Figure 3 is a representation of the heart of an exemplary embodiment of the type "hardware"; it consists of necessary computer cards arranged for proper operation, memories dedicated to databases and useful input / output ports. The remainder of this document is with reference to these drawings. [025] In order to facilitate the understanding of this document: a) The terms "source" and "source of information" should be included in this document as generalizable to any geo-location system regardless of internal techniques and external used upstream or downstream of the calculator of said source, insofar as this system comprises a task for determining the geo-localized position of a part or all of said geolocation system, or of a third party. (b) The term "date" as used in this document refers to time information that distinguishes a "date" from another "date" with a precision better than one-tenth of a second. c) The term "out of service" is used to qualify a task that has declared itself out of service or has been declared out of service by another task as a result of out of bounds delays or inconsistencies detected. d) The terms "reliable" and "unreliable" are used to simultaneously qualify a source and its dedicated Sequence Extraction Task as well as the information they transmit, which qualifiers are awarded following calculations of comparisons of such information. Points [031], [032], [033] explain how these qualifiers are assigned. e) When linked to an element of a list of n elements, x will represent an index corresponding to the order number of the element in the list. The source of index x (1x) is therefore the source of which x is its serial number in the list of sources; (3x) is the Sequence Extraction Task of index x assigned to the index source x (1x). The index port x (2x) is the communication port on which the index source x (1x) outputs information for the Sequence Extraction Task of index x (3x). f) The caps are expressed by their equivalences in radians. (g) The term "bow" shall be understood to mean "small bow". (h) The symbol PK means Kilometric Point. It is a curvilinear abscissa in a one-dimensional curvilinear landmark, corresponding to the course of the train on the track in the direction of movement of the train. This marker has an origin point named "referent", which is often worth 0.000, and a curvilinear unit vector that is usually 1000 meters long. The referent and vector-curvilinear unit is named "referential". The PK is generally expressed in kilometers (example: PK12) followed by a fraction MF (example: PK12.0), if it is necessary of a fraction decametric (example: PK12.05) even of a fraction metric (example: PK12.055). A PK is linked to its reference frame and therefore to its referent (example: PK12.055PAZ, the PAZ trigram defining the referent located at PK0.000PAZ in Paris-Austerlitz). (i) The term "rail" used to qualify a distance or an abscissa means a track length equivalent to that measured at that track. [026] The present method is characterized in that it comprises the following elements: Sequence Extraction Tasks (3x) which transmit information from sources (1x) to an Analysis Task (4) which makes them evaluates and transmits a synthesis to the PKTRAIN task (6) which determines the curvilinear abscissa of the head of the train 20 and transmits it to the following tasks related to the HMI (20): MIRLITONS (7) determines the train-signal distance; EVEIL (9) alert of imminent braking; TOPFREINAGE (10) invites to brake and adapts the process; FOLLOWING (11) assists braking; VITESSEOPTI (12) determines the optimal speed. [027] At a frequency f each source provides the method with Longitude, Latitude, Cap, Speed information characterizing a fixed point relative to the forward end of the train by a known distance from that end. Frequency f is important because it defines the maximum allowable period that can elapse between receiving information from a source and determining the PK of the train head by the PKTRAIN task (6). The execution frequencies allocated to the intermediate tasks result from this. In one embodiment of the present invention, f is equal to 1 Hz. [028] Delivering information to a safety operator such as a train conductor 35 requires that the information be accurate in terms of value and be exploitable at the moment the operator perceives it. This makes it necessary to have control over the intermediate delays of the process since receiving data from the sources until the information is delivered to the operator. To do this, the present invention has a structure consisting of tasks operating simultaneously asynchronously. Each task of this invention performs its steps for a period of its own, a period termed a cycle. Thus at each of its cycles, each task performs the following steps: stores the start date of said cycle; if it uses information from another process task, then it compares the end date of the other task with its own cycle start date, if the result of this comparison exceeds a threshold then it declares itself out service and declares that said other task is out of service and declares that said information is unusable; memorizes the end date of said cycle; compare its own end-of-cycle date with its own start date, if the result of this comparison exceeds a threshold then it goes out of service. In one embodiment of the present invention, the tasks declared out of service retain this qualification for several cycles. [029] The present invention involves a phase of determining a first estimate of the information of longitude, latitude, speed, heading, from the plurality of sources used (1a) (1b) ... (1x) ... ( 1n). This phase requires the implementation of several methods operating in parallel and in real time: Sequence Extraction Tasks (3a) (3b) ... (3x) ... (3n) and the Analysis Task ( 4). There are as many Sequence Extraction Tasks as there are information sources, with each Index Sequence Extraction Task x (3x) assigned an index information source x (1x) and at each index information source x (1x) corresponds to a Sequence Extraction Task with index x (3x). Thus, during each clean cycle of a Sequence Extraction Task of index x (3x), said task performs the following steps: extract from the data stream transmitted by the index source x (1x) the longitude, latitude information , heading, speed; corrects the longitude and latitude information it has retrieved, corrections made based on process execution times. The objective is to anticipate the position of the head of the train corresponding to the future instant when the driver via the HMI task (20) will perceive information computed by the present method following the processing of the data delivered in real time by the index source - 13 - x (1x). This anticipation relates to the period P1 calculated by the HMI task (20) at the [106.c] point during its last completed cycle, and corresponds to the longest of the periods which has been observed by the HMI task (20) between the date of beginning of a Sequence Extraction Task on one or more of its previous cycles and the date of execution of a specific step of the HMI task (20). In one embodiment of the present invention, the Sequence Extraction Task extracts from the data stream information related to the calculation method used by said source and its reliability, and stores them. In one embodiment of the present invention, the Sequence Extraction Task cycles through a period corresponding to a frequency greater than 2f. In one embodiment of the present invention, to effect the necessary correction, the Sequence Extraction Task of index x (3x) determines a geocurvilinear velocity vector - / -; using the value of heading extracted on its current cycle (heading noted in the formulas below) and the value of speed extracted on its current cycle (speed value noted nen in the formulas below). The Sequence Extraction Task of index x (3x) then applies with a period P1 on the geographic coordinates Longitude (noted below) and Latitude (noted below (p) extracted during this cycle. formulas used are: 02 = Arcsinus [cos 0 - sin (liell.P1) - cos (I) + cos M) sin A.1 R cos (lielizi.P1) sin esin 02) A2 = + Arccosinus cos ecos 02 where R is the terrestrial radius at the latitude of the train, and the choice between the + or - noted above ± depends on the sign of sin O. The result consists of a new Longitude noted above A2 and a new Latitude noted above D2. [30] During each cycle specific to the Analysis Task (4), if said task is not out of order then it performs the steps described in [031] to [035]. [31] During said cycle specific to the Analysis Task (4), each Sequence Extraction Task which satisfies the following requirements: - it has not been declared out of service for its last two cycles, - the information that it has extracted evolved between its last two cycles, 35 must be studied by the Task Analysis (4). The study consists of vector comparisons. The Analysis Task (4) then performs the following steps: a) It calculates the cap and modulus elements of the velocity vector which made it possible to connect the point determined by said Sequence Extraction Task at its last cycle from the point determined by said Sequence Extraction Task during its cycle preceding its last cycle. b) It compares by vector subtraction this vector to each of the two velocity vectors determined by each other Sequence Extraction Task declared reliable for its last two cycles: - The first velocity vector is built with the cap and module values speed determined by said other Sequence Extraction Task during its last cycle, - The second velocity vector is the one which made it possible to connect the point determined by said other Sequence Extraction Task during its cycle which precedes its last one. cycle to the point determined by said other Sequence Extraction Task during its last cycle. c) It compares each vector subtraction result to a threshold. d) If said threshold is exceeded then the Analysis Task (4) increments an accumulator related to the Sequence Extraction Task studied by a negative value, otherwise the Analysis Task (4) increments said accumulator related to the Sequence Extraction task studied for a positive value.

In one embodiment of the present invention, the Analysis Task (4) cycles for a period corresponding to a frequency of 4f. In one embodiment of the present invention, the increment values mentioned in [031.d] are calculated based on the precision dilution values (DOP) of the sources and the number of satellites received. [032] A number of sources may emit inconsistent information briefly and almost at the same time, regardless of the number of reliable sources and the number of unreliable sources, with the extreme case where for n existing sources the information from n-1 unreliable sources would be sufficiently concordant with each other to invalidate together the only reliable source remaining. In such a case, the risk would be that the sole source giving reliable information is considered unreliable by the Analysis Task (4) which uses as comparison references the n-1 other faulty sources which would be validated. To solve such problems, each Sequence Extraction Task that meets the following requirements: - it has not been declared out of service for its last two cycles, - 15 - - the information it has extracted has evolved between its last two cycles, must be studied again by the Task of Analysis (4). The study consists of vector comparisons. For each Sequence Extraction Task studied, the Analysis Task (4) performs the following steps: a) it calculates the cap and modulus elements of the velocity vector which made it possible to connect the point determined by the Extraction Task of Sequences during its last cycle from the point determined by said Sequence Extraction Task during its cycle preceding its last cycle, b) it compares by vector subtraction this vector with the two velocity vectors characterizing a so-called virtual source determined by the Analysis Task (4) and the PKTRAIN task (6) during their previous cycles: the first velocity vector is constructed with the heading and module values determined for the virtual source during a preceding task cycle Analysis (4), the second velocity vector is the one that made it possible to connect the determined point P 'for this virtual source during the cycle of the PKTRAIN (6) task which precedes its last one. cycle, to the point P 'determined for this virtual source during the last cycle of the PKTRAIN task (6). c) It compares each vector subtraction result to a threshold. d) If this threshold is exceeded then the Analysis Task (4) increments said accumulator related to the Sequential Extraction Task studied by a negative value, this value not exceeding in absolute value the absolute value of the increment. maximum negative that could be issued by the Analysis Task (4) for a Sequence Extraction Task; e) if this threshold is not exceeded then the Analysis Task (4) increments said accumulator by a positive value, this value being greater than the sum of the maximum increments that the Analysis Task (4) can emit for n-1 Sequence Extraction Tasks. [0331 When all the comparisons have been made, for each Sequence Extraction Task studied, the Analysis Task (4) establishes that: concerning the sum contained in the accumulator related to said Sequence Extraction Task, if said sum is negative then said Sequence Extraction Task was unreliable during its last cycle, otherwise said Sequence Extraction Task was reliable during its last cycle. In both cases, said accumulator is reset. [034] Thus, the Analysis Task (4) evaluates each Sequence Extraction Task according to the following steps: makes vector comparisons between the velocity vectors defined by said Sequence Extraction Task and the vector vectors. speeds defined by a plurality of Sequence Extraction Tasks; assigns a "reliable" characteristic or "unreliable" characteristic to said Sequence Extraction Task based on the results of these vector comparisons. [035] With the information extracted by the Sequence Extraction Tasks declared reliable by the Analysis Task (4) during said cycle, the Analysis Task (4) calculates a P-centroid and a velocity vector using the information transmitted by the Sequence Extraction Tasks declared reliable. In one embodiment of the present invention, the coordinates of the point P are used to characterize the virtual source until PKTRAIN (6) estimates more precise coordinates to define the coordinates of the head of the train. [036] In one embodiment of the present invention, for the same PK defining the curvilinear abscissa of several channels, a coordinate pair (longitude, latitude) per channel is defined in the database 20 VOIES (5). This requires informing the system in an automated or non-automated way of the path taken in real time by the train. In one embodiment of the present invention the number of dimensions that define a PK is an important feature of said PK. Indeed, on some lines traveled, when the lateral deviations between a multitude of channels are important, it is chosen to define the PK either by an affine function or by a polynomial function, functions whose representative curves are intersecting said channels at the abscissa. curvilinear value of said PK. [037] The present method imposes to define in the VOIES database (5) several types of Kilometric Points for the same point located on the route taken by the train. A Real PK is a value of PK which corresponds to the curvilinear abscissa measured at the rail relative to the reference point provided by the document Technical Information of the line. For example, the bumpers of the Paris Austerlitz station are a reference value PK0.000PAZ. A certain point situated in the station of Juvisy-sur-Orge is at the curvilinear abscissa 19,000 compared to this referent, that is to say 19 kilometers measured precisely to the rail compared to the Paris Austerlitz bumpers. This point located in Juvisy-sur-Orge station has as PK Real the PK19PAZ. Juvisy-sur-Orge being the junction of several lines, this same point also has another value of PK Real in the repository of another line. A PK Terrain is a PK value defined by infrastructure elements present in the field, such as PK inscriptions carried on a physical terminal or on a catenary pole located along the track. The implementation of some of these elements may vary depending on various events such as for example following work on the tracks. There are cases of PK adjustment where a terminal carrying a certain value of PK is followed a few tens of meters further from another terminal carrying the same value of PK. Updating PK Fields in the database requires frequent adjustments to be in tune with the reality of the field.

The referee of a PK Terrain is the same as that of the PK Real which corresponds to him. The Virtual PK are to be compared to the Real PK in that they want to be precise in terms of curvilinear abscissa measured with the rail compared to a referent. They are called "virtual" because they can be linked to different repositories from that of the Real PK where is the head of the train. Thus the origin point of a train, regardless of its Real PK, can be used as a referent of Virtual PK. If one or more line transitions are present on the route, some calculations are simplified because they are made taking into account a single repository valid throughout the route. To reach the same station there may be several routes of different lengths, all the possible routes for the same train are therefore provided in the database, 25 which makes it easy to manage the cases of modified routes en route. The use of PK Virtual Correspondence Tables - 'PK Terrain' makes this technique completely transparent to the train driver. In one embodiment of the present invention, since the modifications of the route taken by the train are decided by a control service, this service has the task of transmitting them to the present method via the data transmission function. included on the Sol-Train Radio (14) via a 3G key (15) or any other means of communication. [038] The plan outline of the route includes straight alignments, circle arcs, and progressive curvature curve arcs. In France, the progressive connection of reference is a clothoid, this one -18- answering generally the necessary conditions on the course to be able to be assimilated to a cubic parabola. Right alignments are treated by the present method as arcs of curves whose curves are straight lines.

The present method manages all the shapes of curves defined by the topography services assigned to the drawing in plan, it also manages all the shapes of curves comparable to one or more curves with known characteristics. [39] The route is cut into contiguous sections so that each section 10 corresponds to a curve arc with characteristics defined by the plane plot, or to a similar curve arc. The information for characterizing each section of the route is stored in the channel database (5). [40] Thus the database VOIES (5) is characterized by the following definitions: (a) at each possible route for the train, there exists in the database VOIES (5) a list containing: i) the characteristics of the reference point of the said list; (ii) geographical points separated by 1000 meters from the rail. For example, the virtual PKs with virtual values that are integer with respect to the referent will be chosen, iii) geographical points origins or ends of arcs of curves defined by the plane plot, iv) a correspondence table necessary to convert the virtual PKs into PK Terrains, b) each geographical point is defined in the database VOIES (5) by: i) its classification index in said list, and in the direction of movement of the train, if said train meets the PK 25.000 then the The origin of a clothoid arch at PK 25.120 then the end of this clothoid arc at PK25.600 followed by the origin of an arc at PK25.600 followed by the end of this arc of circle at PK25.800 followed by the origin of an arc of a second clothoid at PK25.930 followed by PK26.000 followed by the end of the arc of this second clothoid at PK26.150, then the list will have PK25.000 of classification index k, followed by PK25.120 of classification index k + 1, su ivi PK25.600 k + 2 ranking index, followed by PK25.800 ranking index k + 3, then PK25.930 ranking index k + 4, then PK26.000 35 classification index k + 5, then the PK26.150 index k + 6 ii) its coordinates, - 19 - iii) its curvilinear abscissa with respect to said referent, iv) the information characterizing the curve arc and the curve on wherein said point is inscribed, information and curves being defined by said plane path. (c) the same point is present only once in the same list except in the case of a crossing or where the point more than once defines the center of that crossing. In one embodiment of the present invention, the TRAIN database (5) assigned to the train is constructed in an automated manner by a tool that processes information from several national databases. [041] From the estimation of the point P carried out by the Analysis Task (4) during its cycle, the task PKTRAIN (6) aims to establish the PK of the head of the train. The objective is to treat the problem of determining the curvilinear abscissa of the head of the train on the track as that of the determination of the curvilinear abscissa of a point P 'situated on a curve C with known characteristics, curve C whose r is an arc. On C, P 'is the point closest to the point P calculated by the task PKTRAIN (4). [042] During each cycle proper to the task PKTRAIN (4), if said task is not out of service during said cycle then the task PKTRAIN (4) begins by determining in the database VOIES (5) the points PK1 and PK2 defining respectively the origin and the end of the curve arc on which is located the head of the train. Classic approaches to this problem often lead to looking for the two closest points to the head of the train, one of which is "past" and the other "to pass". The problem that arises then is that the geographical configurations of certain lines lead, during the same route, a train to enter the same geographic area several times. If upstream of a first step in said sector the sources are faulty and they become again usable in said sector, according to the manufacturer-margins of said sources the risk is high that a method using said conventional approach does not know which points recorded in the database correspond to the current step. The encounter of such a problem is all the more likely if, during the various stages, the paths taken intersect or approach each other in this sector by a distance less than the manufacturer's margins of said sources. Note that in such a case the identification of a portion of the route according to a heading would require that the tracks are distinguished by the caps of their direction of movement, which can lead to errors if these course values are too close to each other or if the sources used give course information deemed exploitable but whose error tends to the threshold of rejection of a path by the method. The solution provided to these problems by the present invention consists in the definitions of the elements contained in VOIES (5) and in the following steps. Note that the virtual PK points flanking the head of the train are to be found in VOIES (5) in the list of points corresponding to the route taken. [43] If the PKTRAIN (6) task was not out of service during its previous cycle and determined the PK of the head of the train, then the PKTRAIN (6) task uses the value of the determined PK during that cycle. PKTRAIN cycle (6), to estimate the two points PK1 and PK2 sought, then it goes to point [052]. Otherwise the task PKTRAIN (6) applies the following points. [44] Let P be the position estimate of the head of the train provided by the Analysis Task (4). In the list of virtual PK points corresponding to the route, the PKTRAIN task 15 (6) searches for the four virtual PK points whose distances to the point P are the shortest and less than or equal to a threshold. In one embodiment of the present invention this threshold is equal to a distance equal to the greater of the distances separating two points to the ranking indices following from 1 in VOIES (5), thus 1000 meters, to which are added manufacturer margins. [45] If zero or a single PK corresponds to the criteria in [044], then the VOIES (5) database has an error. The PKTRAIN task (6) is then declared out of service during this cycle. [46] If two PK points correspond to the criteria of [044] then the PKTRAIN (6) task verifies that the difference between their classification indices in the list is 1. If this is not the case, then (5) has an error, PKTRAIN (6) is then declared out of service during this cycle. Otherwise the task PKTRAIN (6) names PK1 30 that of said two points PK which has the lowest classification index in said list and PK2 the other point PK. The process found the two desired points, then PKTRAIN (6) goes to [052]. This case is the one most frequently encountered. [47] If three PK points correspond to the criteria of [044], then the task PKTRAIN (6) looks for which of these three PK points have successive rankings in that list with a difference of 1. -21 - a) If no PK point has this feature, then CHANNEL (5) has an error. The PKTRAIN task (6) is then declared out of service during this cycle. b) If two PK points have this characteristic then PKTRAIN (6) names PK1 that of said two points PK which has the lowest classification index in said list and PK2 the other point PK. The process found the two desired points, then PKTRAIN (6) goes to [052]. This case may be encountered if the lane has sections that approach or intersect. c) Otherwise PKTRAIN (6) defines the two bipoints formed by these three PK points such that the two PK points of each of these bipoints have a difference of 1 between their classification indices in said list. PKTRAIN (6) then goes to [049]. This case occurs if the train is near a PK point recorded in VOIES (5) and the error margins of the sources lead to significant errors in their lateral positioning with respect to the movement. [048] If four PK points correspond to the criteria of [044], then the PKTRAIN (6) task defines the bipoints formed by these four PK points such that the two PK points of each of these bipoints have a difference of 1 between their two points. classification indices in the said list. a) If no PK point has this feature, then CHANNELS (5) has an error.

The PKTRAIN task (6) is then declared out of service during this cycle. b) If it turns out that only two of these points PK form a bipoint, then PKTRAIN (6) names PK1 that of said two points PK which has the lowest classification index in said list and PK2 the other point PK. PKTRAIN (6) then found the two desired points, then PKTRAIN (6) goes to [052]. This case may be encountered if the lane has sections that come closer or cross more than once. c) If it turns out that only three of these four points PK form bipoints, then PKTRAIN (6) defines the two bipoints formed by these three points PK and PKTRAIN (6) then passes to the point [049]. This case is similar to point [047.c], but with the additional feature that the track has sections that approach or intersect with each other. d) If it turns out that four PK points form bipoints and if the classification indices of the four PK points follow each other with a difference of 1 in the list then three bipoints are formed and PKTRAIN (6) goes to [049 ].

But if the ranking indices of the four PK points follow each other two by two with classification index differences of 1 in said list, then PKTRAIN (6) -22- defines the two corresponding bipoints and goes to the point [049]. This case occurs when the lane has sections that approach or intersect; this case will be encountered more frequently if the error margins of the sources lead to significant errors in their lateral positioning with respect to the movement. [049] For each bipoint, the PKTRAIN (6) task names PK1 the one of the two PK points which has the lowest classification index in the list, and PK2 the other PK point. If one of these two points is neither an origin nor a curve arc end 10 provided in VOIES (5), then PK1 takes the characteristics of the origin point of the curve arc on which these two points inscribe, and PK2 takes the characteristics of the end point of the curve arc on which these two points are inscribed. considering that PK1 and PK2 are respectively the origin and the end point of an arc r of a curve C, r and C are characterized by the plan line; 15 considering that there exists a point P 'of the curve C, such that P' is the point of C closest to the point P, the task PKTRAIN (6) performs the following steps: a) it calculates the normal to C passing through P 'and P, normal whose directional vector is -1 / F', F 'being the derivative at P' of the function F, of which C is the representative curve, b) it deduces therefrom the coordinates of the point P 'using if necessary an algorithm such as Newton's method. The point found is named P '. The orthodromic distance d (Bipoint) between P and P 'is calculated and associated with the studied bipoint. For each bipoint, the PKTRAIN task (6) stores the coordinates of the point P 'to avoid recalculating them at the point [053] of the present method. [050] If PKTRAIN (6) is in the case of four PK points forming bipoints two by two, then PKTRAIN (6) compares their values d (Bipoint) (ATTENTION: these values 30 are not the distances between the two points of each bipoint, but the distances between P and P '). If the difference between these values is below a threshold, then PKTRAIN (6) is declared out of service during this cycle. This case is met if the train is near a crossing where it will pass on the track located in a certain direction and later on the track in the other direction. Because of a concern for accuracy related to the sources, on its current cycle the PKTRAIN (6) task can not determine on which of the two paths the train is located, so it waits for at least one cycle as the situation evolves. - 23 - [51] If the task PKTRAIN (6) is not in the situation described in [050] then said task looks for the bipoint to which is assigned the great circle distance d (Bipoint) the smallest of all the distances d (Bipoint) calculated during this cycle. The PKTRAIN task (6) names PK1 the PK point of this bipoint which has the lowest ranking index in the list and PK2 the other PK point. [52] By controlling the times of each task of the present invention, on the portion of the path between PK2 and PK2 + E with -E tending to 0, the shape of the pathway beyond PK2 tends to the shape that the path possesses between PK1 and PK2. This feature makes it possible to consider a point P 'located slightly downstream of the arc [PK1, PK2] as being a point of the same curve, but under two conditions: firstly that the Analysis Task (4) and the PKTRAIN task (6) not considered out-of-service at this time, which is the case if PKTRAIN (6) arrives at this point in the process; secondly, P 'is close to PK2. To satisfy this second condition, it must be understood that the notion of curvature of track and that of limit speed of traffic on the track are related, which leads to perceive the notion of distance between the point P 'and PK2 in terms of time. As in the previous cycle of the PKTRAIN task (6) the point P 'calculated on said previous cycle was located on the arc [PK1, PK2], then the distance between PK2 and the point P' calculated on the current cycle of PKTRAIN ( 6) is, expressed in terms of travel time, less than 1 / f when P 'is beyond PK2. (see point [027]). Without the study of this peculiarity the good functioning of PKTRAIN (6) would be impossible since this task would behave indefinitely as if the point P 'was on the arc [PK1, PK2] and not beyond PK2, and the PK value of the P 'point would never evolve beyond the PK2 PK2 value. [53] If one of these two points PK1 or PK2 is neither an origin nor a curve arc end provided in VOIES (5), then PK1 takes the characteristics of the point of origin of the curve arc on where these two points are inscribed, and PK2 takes the characteristics of the end point of the curve arc on which these two points are inscribed. Knowing PK1 and PK2, if it has not already done so, the task PKTRAIN (6) can now calculate an estimate of the coordinates of the head of the train on the curve C whose r is an arc.

Considering that PK1 and PK2 are respectively the origin and the end point of an arc r of a curve C, r and C characterized by the plane line; - 24 - considering that there is a point P 'of the curve C, such that P' is the point of C closest to the point P, the task PKTRAIN (6) performs the following steps: a) it calculates the normal to C passing through P 'and P, normal whose direction vector is -1 / F', F 'being the derivative at P' of the function F, of which C is the representative curve, b) it deduces from it the coordinates of the point P 'using if necessary an algorithm such as Newton's method. P 'is a better estimate of the longitude and latitude of the head of the train than P. 10 The task PKTRAIN (6) validates it thus as that of the head of the train, and as element characterizing the virtual source that the Task Analysis (4) will use during his next cycle. [54] Task PKTRAIN (6) computes the curvilinear abscissa of P 'with respect to the origin point PK1, and adds it to the value of the Virtual PK of PK1. The virtual PK of the head of the train relative to the referent of said list takes the value of the result of this calculation. [55] The PKTRAIN task (6) estimates the PK Terrain of the head of the train using the correspondence table "PK Virtual - PK Fields". [56] Thus the PKTRAIN task (6) performs the following steps: determines the point PK1 origin of an arc r of a curve C, and the point PK2 end of said arc r of said curve C, the head of the train being situated on said curve C, said curve C and said arc r being defined by the plane plot; Determines the curve C passing through PK1 and PK2; determines the normal to the curve C, normal passing through P; determines the coordinates of the point P 'of the curve C closest to P; determines the curvilinear abscissa of P 'on the curve C; defines that the coordinates of the head of the train are those of P '; 30 determines the abscissa curvilinear to the rail of the head of the train and deduces its PK. [57] In one embodiment of the present invention, the HMI task (20) displays with two decimal places the PK Terrain of the head of the train as well as the reference of the PK Terrain (example: 19.15PAZ), this display being done on a screen in front of the driver. The PK Terrain information available in real time for the driver helps to solve problems of misrepresentation of the positioning of the train, sequence break, misclassification of actions, problems related to a need to locate the position of the head of the train on the track, to identify the position of an obstacle. In one embodiment of the present invention, a third party system provides the PKTRAIN task (6) with distance information to the rail relative to stationary equipment located along the path, the PKTRAIN task (6) having access to PK data of such equipment and using said information and said data to calculate the PK of the head of the train. [058] For each possible route of the train, the SIGNAUX database (8) comprises for each signal encountered: a) its virtual PK, b) its identification information (name, shape of its target, identification plate channel concerned) It may also include the information that said signal displays according to the route. [59] Knowing the PK of the head of the train, the task MIRLITONS (7) uses the information of the SIGNAUX database (8) to calculate the distance to the rail between the head of the train and the next signal encountered. Such information disseminated to the driver helps to solve problems of poor positioning with respect to a signal, and certain problems of sequence breaks leading to stop signal crossings. In one embodiment of the present invention MIRLITONS (7) indicates to the driver, via the HMI task (20), bearings of distances which separate the head of the train 25 from this signal (for example: 800m, 600m, 500m, 400m, 300m , 200m, 100m, 50m). In one embodiment of the present invention this aid consists of brief audible alerts indicating these distances when the corresponding steps are crossed and when they concern a closed square signal or a closed semaphore. In one embodiment of the present invention a graphical representation of the plane of the tracks and signals with respect to the position of the train is also provided, a representation scrolling in real time and indicating which signals have been crossed, which ones to come, which signaling must be met. [60] Thus the task MIRLITONS (7) estimates the distances to the rail between the head of the train and 35 signaling elements present on the route. [61] The GARES database (13) contains PK information of each planned stopping point according to the route, mission, and train composition. [62] At each of its cycles, the task EVEIL (9) calculates the braking distance DiNmALE necessary to stop from an initial velocity VINITIALE taking into account various parameters of the train. It then compares DINITIALE with the distance to the rail Dl which separates the head of the train from the planned stopping point, to which are added various distances representing, according to VI, NITIALEI several time margins. When the result of this comparison is below a threshold then the alert is broadcast. DINITIALE is expressed in terms of VINITIALE in DINITIALE = F (VINITIALE, (PI A, w, L, p) where w is the weighted average gradient on the course corresponding to the braking, L the length of the train, A: a formless coefficient representing the percentage of mass braked with respect to the mass of the train, cP: a braking coefficient given by tables, and p: a factor related to controlled braking During the first braking performed or when it is necessary, EVEIL (9) gives p a value representing a first service depression, the task EVEIL (9) thus contributes to correcting the omissions of stops which follow sequence breaks occurring before the abscissa corresponding to DINITIALE, or which cause due to a poor representation of the position of the train or the stopping point, or even to a poor representation of the braking capacity of the train, in one embodiment of the present invention, DINITIALE can be expressed r in the form: DINITIALE = VINITIALE2 - G "y, A, CO, L, In one embodiment of the present invention, the distances representing the added time margins are calculated as a function of: the broadcast alert message; periods of human perception, periods of analysis and reactions on the brake control, reaction times and balancing of the brake equipment (variable depending on the characteristics of the train), a minimum period characterizing the phase time arousal. In one embodiment of the present invention the alert is sound: "Prepare to brake." In one embodiment of the present invention, EVEIL (9) is also responsible for detecting a stop made in the open lane and for controlling the display on a screen of various turns of hands related to settlement and resumption of operation. . - 27 - [63] Thus during each cycle specific to the task EVEIL (9), the task EVEIL (9) carries out the following steps: estimates D1: the distance to the rail between the head of the train and the next planned stopping point ; estimates the DINITIAL braking distance required for the train to stop; issues an alert if the difference between D1 and DINITIALE is less than a threshold. [64] The task TOPRREINAGE (10) compares DINITIALE with the distance to the rail which separates the head of the train from the planned stopping point to which are added various distances representing, according to VINITIALE) several periods. When the result of this comparison is less than a second threshold then an alert is broadcast. Thus TOPFREINAGE (10) helps to remedy the omissions of stops that occur after breaks in sequence occurred between the date of issue of the waking information and the braking point.

In one embodiment of the present invention, the distances representing the added time margins are calculated according to: the alert broadcast, periods of human perception, analysis and reactions on the brake control, times of reaction and balancing of brake equipment (variable depending on the characteristics of the train). In one embodiment of the present invention the alert is sound: "brake". [65] During braking, the TOPFREINAGE task (10) compares with a third threshold the distance to the rail which separates the head of the train from the planned stopping point. If said threshold is crossed then an alert is broadcast. Thus TOPFREINAGE (10) helps to remedy the omissions of stops that follow breaks in sequence occurred between the beginning of the braking and the critical abscissa separating the rail head of the train from the stopping point. In one embodiment of the present invention, the threshold represents the length of the train to which various margins are added. In one embodiment of the present invention, the alert is soundly broadcast: "Stop" followed by the name of the marker on the ground and concerning the characteristics of the train. [066] During the first braking performed to stop at a planned stopping point, the tasks EVEIL (9) and TOPFREINAGE (10) are content to give p a theoretical value representing a first depression of service, that is to say an early braking and more accentuated. But during the following braking, the tasks TOPFREINAGE (10) and EVEIL (9) adapt to the performance of the couple man-machine to deliver their alerts at the optimal moments for this couple. To do this, the TOPFREINAGE task (10) has information on the behavior of the man-machine torque at each braking performed and its evolution. Thus, at each braking performed to perform a planned stop, at each of its cycles, the task TOPFREINAGE (10) collects in real time the curvilinear abscissa information Xi of the train with respect to the next scheduled stopping point and the information of Vi velocity delivered by the Analysis Task (4). Xi is the result of the subtraction between the Virtual PK of the head of the train and the Virtual PK of the stopping point. i is an index incremented at each collection performed and reset at the beginning of each braking. Once the shutdown is detected, the TOPFREINAGE (10) task processes this information as explained below. [067] For the harvested data to be exploitable, the corrections made during braking by the driver via the brake control must be limited in number and amplitude. In one embodiment of the present invention, it is defined that this condition is mainly achieved when the speed of the train is greater than 60 km / h. Indeed, it is below this speed that the driver acts most frequently on the brake control to try to refine its braking and stop precisely. [68] The minimum number of points collected during braking is set at a threshold s.

In one embodiment of the present invention, a threshold is set at 12 which gives excellent results with any rolling stock. [69] The braking performed must correspond to a "standard" braking to stop. Any braking that begins after the waking information is transmitted and ends with the stopping of the train near the planned stopping point is considered a "typical" braking. [70] A minimum correlation rate of the curve obtained with respect to the cloud of pairs (Xi, Vi) taken is imposed. In one embodiment of the present invention, this minimum correlation rate is set at 98%, a condition which is satisfied in the vast majority of cases when sources of general public geolocation are used. - 29 - [71] The process must take into account the type of conduct performed. In one embodiment of the present invention, for a suburban train of type Z2N, the pair (200; 60) is added to the cloud of collected pairs, this to indicate the wish to be at 60 km / h when the train will be 200 meters from the stopping point. Included, the couple (0; 0) mark that the goal is to be at 0 Km / h at the planned stopping point. [72] Xi = F (Vl, (p, A, w, L, p) can be expressed in the form Vi = b - H (Xi), so that when the train was stopped following a typical braking to stop at a breakpoint identified in the GARES database (13), if the thresholds related to the pair cloud (Xi, VI) and defined in points [067], [068], [069 ], [070], [071] are followed, then the TOPFREINAGE task (10) uses the pair cloud (Xi; Vi) to establish the y intercept, the b coefficient, and the correlation coefficient c, of the curve of least squares of form: Vi = a + b - H (Xi).

In one embodiment of the present invention, H (Xi) =, 1 Xi [73] If c is greater than the threshold defined as the minimum correlation coefficient, then the TOPFREINAGE task (10) is allowed to use b and a for estimate the value of p. Thus, during the next braking to be performed to stop at a planned stopping point, the tasks EVEIL (9) and TOPFREINAGE (10) will apply the new value of p in the formula D INITIALE = F INITIALE) (1), A, w, L, p). The values of b and a can then be used by the VITESSEOPTI task (12) described below. [74] TOPFREINAGE (10) is connected to a position detector of the brake control on the control panel (16a), to a manometer (16b) measuring the air pressure in the General Conduct of the train, and, following the engine used, a computer type ATESS (16c) managing the distribution of the braking force between the various organs and devices to obtain it. When braking causes TOPRREINAGE (10) to detect inconsistencies between the values provided by these devices and braking curves made to stop at the points provided, then TOPFREINAGE (10) issues an alert to technical services. In one embodiment of the present invention this alert is via the Sol-Train Radio (14) or via a 3G key (15). In one embodiment of the present invention the actual braking degree information transmitted to the process may consist of an electric current or air pressure or the position of a brake member or be provided information. - the process by an embedded system. [75] Thus during each cycle proper to the task TOPRREINAGE (10), the task TOPRREINAGE (10) performs the following steps: emits an alert when is below a threshold the difference between D1 and DINITIALE emits an alert when the distance D1 is below a threshold; stores each pair (Xi; Vi), i being an index incremented at each storage and reset at each braking, Xi being for a value of index i the distance to the rail separating the head of the train from the planned stopping point, and Vi being for a value of index i the speed of the train recorded at abscissa Xi; calculates the intercept a, the director coefficient b, and the correlation coefficient c, of the shape least squares curve: Vi = a + b - H (Xi); authorizes the task EVEIL (9) and the task TOPFREINAGE (10) to be used during the next braking the calculated results; compares the degree of braking performed with the degree of braking commanded and issues an alert if the result of this comparison exceeds a threshold. [76] Tests carried out on 220900 self-propelled equipment have shown that assisted braking with the present method makes it possible to carry out what is known in the jargon as "shots on purpose", that is to say stopping with a single brake stroke without correction until it stops at the desired point. An embodiment of the present invention provides not to exploit the newly calculated parameters a and b by TOPFREINAGE (10) if the time elapsed between the date of issuance of the alert provided for in [064] and the date of issue. obtaining the corresponding stop is less than the theoretical delay to achieve a complete emergency braking of pneumatic type that is to say without use of electric braking or use of a braking type electromagnetic. In one embodiment of the present invention, coupled to the SIGNAUX database (8), the MIRLITONS task (7) performs on each of its cycles calculations equivalent to those of the points [072] and [073] to detect a braking which aims to stop in front of a stop signal. [77] Monitoring the deceleration of a train in braking mode to stop at platform 35 makes it possible to indicate to the driver whether he should increase or decrease the intensity of the braking in progress with a view to stopping at the scheduled stopping point. During said braking, by using some of the very last pairs (Xi; V;) recorded in real time by TOPFREINAGE (10), the FOLLOWING task (11) determines the leading coefficient b 'of the least squares curve. J (x) = a '+ b' - H (x). [078] The FOLLOWING task (11) then calculates the derivative of the function J at the abscissa X of the train, abscissa to the rail relative to the planned stopping point, and deduces therefrom the angle a-real of the tangent to J at P2 (X, J (X)). The theoretical point P2 located on the curve representative of J is the point on this curve which should theoretically correspond to the speed of the train as a function of: its X abscissa to the rail relative to the stopping point, the braking power achieved in real time. In one embodiment, H (X) = JX, hence J (X) = a '+ b' - Ni X, so the derivative of J in P2 (X; J (X)) is equal to b '/ ( 2 -N X). [079] The FOLLOWING task (11) then determines the braking coefficient b "that the man-machine torque should theoretically achieve at the point P3 (X; V), X being the abscissa at the rail of the train relative to the point of In one embodiment of the present invention, b "= V" NI X [080] SUIVIFREINAGE (11) derives the a-theoretical angle from the P3 (X; V) of the representative theoretical curve of L such that L (X) = b "- H (X) In one embodiment of the present invention, the directing coefficient of the theoretical curve of L at P3 (X; V) , such that L (X) = b "- -NI X, is equal to b" / (2 JX). [081] The angle a-real is the angle realized in real time by the man-machine pair, and the The a-theoretical angle is one that should be made at the same place at the actual speed of the train in order to stop at the planned stopping point. [082] With A = a-theoretical - a-real, A allows quantify the correc necessary to perform on the brake control, correction information transmitted by the FOLLOW-UP task (11) by means of the HMI task (20). The FOLLOW-UP task (11) thus contributes to solving the omissions of stops and the bad execution of stops, problems that can occur at any time following a break in sequence, a poor representation of the train on the network, a poor control of the brake control, confusion on the characteristics of the train. - 32 - [83] Thus for a braking in progress the FOLLOWING task (11) estimates the correction to be made on the brake control. [84] The objective of the following points is that the VITESSEOPTI task (12) calculates in real time the optimal speed at which to roll in order to arrive on time at the planned stopping point, taking into account the speed transitions. which mark the remaining distance to the stopping point, the length of the train and the performance of the train in terms of acceleration and braking. VITESSEOPTI (12) operates a database called VITESSESVOIES (19) updated when necessary. a) SPEEDWAYS (19) contains the Virtual PKs of the entry and exit points of the transition speed zones of the course, and the corresponding speeds. These areas may be permanent or temporary, or may include construction sites requiring temporary lowering speeds. b) When necessary, this database is temporarily modified by the addition of one or more Cautious Walk Zones. According to various embodiments, the braking curve used may have undergone a modification to generate anticipated braking, and the acceleration curve may also have undergone a modification to generate a delayed acceleration, anticipation and delay calculated as a function of periods relating to different margins of human safety and parameters related to the train: human reaction time, reaction time and balancing of brake equipment and so on. However, for the sake of simplification of the present document, in the points below, it is considered that no such modification is made to these curves. According to various embodiments, the addition of safe walking zones can be carried out by the train driver or via the Sol-Train Radio (14) or via a 3G key (15) by transmitting their characteristics by an agent in charge of the train. hand over the Prudent Steps to the driver. The driver is then responsible for making the necessary corrections and additions. The transmission can also be done by means of a USB key (17), or by any other means of data transmission. [85] Each zone to be treated is numbered, Z being the index that defines the sequence number. Each zone to be treated comprises an initial abscissa Xiz, a limit speed VLz, an abscissa of end of zone Xfz. So Xiz = Xfz_i and Xfz = Xizo. [086] If Vx is the maximum speed that can be achieved legally and technically, then Vx must not exceed the VLz of that zone. - 33 - [087] The entry of the train in zone of index Z corresponds to one of the following cases: either after braking to reach Vx, or following a zone of index Z-1 where VLz_i <Vx, in this case, we will try to accelerate towards Vx once the tail of the train has crossed Xiz, either to continue a braking or an acceleration that has already started in zone of index Zn and ending in zone of index Z + p (n > 0 and p> 0), which may be the case if the length of the train is greater than that of one or more zones lying between the zone of index Zn and the zone of index Z + p. [088] The points [089], [090], [091], [092], [093], make it possible to create a BDE Ephemeral Database (18) comprising the achievable speed on each fraction of the route which remains at train to go until his next scheduled stop. In one embodiment of the present invention, the course fractions have a length of one meter, a value which makes it possible to obtain satisfactory results. [89] For each zone of index Z, starting from the last zone to be met by the train to reach the planned stop and going up to the zone following the one where the head of the train is located, if VLz> VLz., Then the task VITESSEOPTI (12) redefines the abscissa Xiz from which the entire train has cleared the beginning of said Z index zone. To do this on said Z index zone, the task VITESSEOPTI (12) defines that if the train is to accelerate then it can only do so from the initial abscissa Xiz extended by a distance L, where L is the length of the train. [90] For each zone Z from the location of the train to the location where its scheduled stop is located, the VITESSEOPTI task (12) assigns the VL of the Z index area to each fraction of the track that the train will travel on the zone of index Z taking into account, if necessary, constraints relating to the length of the train established in point [089]. [91] For each zone of index Z from the one following the one in which train 30 is located to that where its planned stop is located, if VLz +, then the train must brake to be at maximum speed VLz .i to the abscissa Xfz. In this case, the VITESSEOPTI task (12) performs the following steps: a) The VITESSEOPTI task (12) relates to Xfz the braking curve resulting from the formula V = b - H (X) where b is a function of the characteristics of the train, by recalibrating said curve so that it takes into account VLz + i as the final speed. If all the necessary conditions have been fulfilled for the task to validate the results of the calculations performed in [072], then b will correspond to the performance of the last braking performed. b) Depending on the braking curve, VITESSEOPTI (12) then defines the speed that the train will have on each fraction of Xfz upstream travel without exceeding the regulatory limit speed of the zone studied. Note that the deceleration curve can extend over several areas. [092] For each zone Z from the one following the one in which the head of the train is situated until the station where it stops, if VLz> VLz., Then the VITESSEOPTI (12) 10 task performs the following steps: a) The VITESSEOPTI task (12) locates the zone clearance abscissa by the tail of the train, the abscissa possibly modified at point [089]. From this abscissa the task VITESSEOPTI (12) applies the acceleration curve, shifting said curve so that it takes VLz ..., as the initial speed. B) The VITESSEOPTI task (12) then defines, on each fraction of the accelerated course, the speed that the train will have without exceeding the VLN of the zone N studied, or to extend on the fractions of courses concerned by a braking treated to the point [091]. Note that the acceleration curve can extend over several areas. [093] Concerning the station stopping point, the VITESSEOPTI task (12) performs the following steps: a) The VITESSEOPTI task (12) applies the braking curve resulting from the formula V = b - H (X) to recaling said curve so that it takes into account 0 km / h as the final speed at the planned stopping point. If all the necessary conditions have been fulfilled for the TOPFREINAGE task (10) to validate the results of the calculations performed in [072], then b will correspond to the performance of the last braking performed. b) Depending on the braking curve, VITESSEOPTI (12) then defines the speed 30 that the train will have on each fraction of travel upstream of the planned stopping point without exceeding the regulatory limit speed of the zone studied. Note that the deceleration curve can extend over several areas. [094] Proceeding in this way makes it possible to guarantee that: on a fraction of the studied course, if a speed resulting from a braking curve and a speed resulting from an acceleration curve are applicable, then the lowest of these two speeds will be imposed - 35 - on this fraction of the course. The guarantee is also obtained that the speed limit attributed to a fraction of the route by VITESSEOPTI (12) is never greater than the limit speed of the zone on which it depends. [095] The database BDE (18) then comprises the speeds achievable by regulation and technically along the entire path that separates the train from its next stop at the quay, with increments functions of the length of the course fractions, taking into account the characteristics of speed zones, acceleration and braking performance of the train, train length.

This information base will be ephemeral because, the train moving at each cycle, it will only be valid during a period of one cycle of the Analysis Task (4). [096] The result of the application of these limit velocity zones and these curves on the different fractions of routes encountered by the train to the next stopping point thus generates a curve consisting of braking, acceleration and limit speed stops. In one embodiment of the present invention, before calculating the optimized speed according to the BDE (18), VITESSEOPTI (12) compresses the contiguous course fractions that are at identical speeds. This reduces the number of calculations performed at the points below. To do this VITESSEOPTI (12) modifies the BDE (18) so that it contains zones of one meter in length on the acceleration or braking portions, and longer zones on the sections where the VLz is constant. [097] Proceeding by dichotomy VITESSEOPTI (12) varies the maximum speed (variable speed noted Vopti) throughout the course studied, respecting the following rule: if in an interval of the BDE (18) Vopti is greater than the VL recorded for this interval, then only for this interval Vopti will be equal to said VL. VITESSEOPTI (12) makes the variations of Vopti by dichotomy starting with half of the maximum speed limit that can be reached on the course and stopping the calculations beyond a certain number of iterations. The calculations performed at points [098], [099], [100] are performed for each value that Vopti takes. The number of iterations necessary to obtain a result to the nearest km / h is calculated as follows: n = [Log2 (Vmax + 1) 1. For a Z2N whose maximum speed is 140 Km / h, it will take 8 iterations, for a TGV that can travel up to 360 Km / h it will take 9 iterations. -36- [98] In an accumulator, VITESSEOPTI (12) notes the time T that the train will have to brake on the portion of the track concerned by the braking treated at [093]: x0 T = fo b - H (- _ _ 'X) - dx where T is expressed in seconds, where b is a function of different train parameters (length, mass, braking weight achieved, grip, type of brake shoes, type of braking, gradient, braking amplitude. ..), x0 is the distance in meters to go from the initial speed VO to 0 m / s. If all the necessary conditions have been fulfilled for the TOPFREINAGE task (10) to validate the results of the calculations performed in [072], then b will correspond to the performance of the last braking performed. [99] In this same accumulator is added if necessary the initial acceleration time to reach the speed not to be exceeded in the area where the train is located. In one embodiment of the present invention, the acceleration curve is expressed as a monomer of degree d, the acceleration time T being then: 1-d 1-d Xf cxd Vr- -VO-d- - dx =, where X0 is the distance from T = ixo (1- (1) .ed 20 0km / h to the initial speed of the train VO, and Xf being the distance to go from 0 km / h at the final velocity Vf, Vf corresponding to VLz if Vopti> VLz, or Vopti if Vopti <VLz [100] For each value of Vopti, a) The VITESSEOPTI task (12) adds to the accumulator every necessary time 25 to travel each fraction of a journey to the stopping point. b) VITESSEOPTI (12) adds the current time to the result of the accumulator. c) If the result obtained under [100.b] is later than the expected arrival time at the stopping point, then the VITESSEOPTI task (12) again performs the points calculations [098], [099], [100] with a higher velocity Vopti, speed estimated by dichotomy as explained in [097]; otherwise the VITESSEOPTI task (12) decreases Vopti by dichotomy as explained in [097] and performs again the calculations of the points [098], [099], [100]. [101] When the expected number of iterations has been performed, VITESSEOPTI (12) stops the calculations made at [098], [099] and [100]. Vopti then has the value of the desired optimal speed. - 37 - [102] VITESSEOPTI (12) then estimates whether the Vopti information can be transmitted to the driver or not. It will not be in the following cases: a) when braking to stop at a planned stopping point is in progress, b) approaching a speed transition zone at a speed below Vopti (c) when Vopti is greater than the limit speed of an area where part of the train is located, and (d) when the schedule at the next stopping point can not be made with a defined margin. An embodiment of the present invention provides the driver with the ability to anticipate or delay the arrival time at the planned stopping point according to the requirements transmitted in real time by the traffic control agent. The process then adapts and provides the optimized speed that results from this adaptation. [103] Thus the database VITESSESVOIES (19) contains PKs of the entry and exit points of the zones of different speeds of the route, as well as the corresponding speed limits. [104] Thus, during each cycle specific to the VITESSEOPTI task (12), the VITESSEOPTI task (12) performs the following steps: for each fraction of the journey until the next scheduled stop, estimates the speed limit that can be technically and legally train; calculates a speed to make the schedule at the next scheduled stop; establishes whether said speed is relevant information according to regulatory and hourly limitations. [105] Line tests reported that the conduct of a Z2N commuter train guided by this method without the occurrence of a particular event made it possible to arrive at the planned stopping point with less than 5 seconds. error in 80% of cases and less than 10 seconds of error in 95% of cases without ever exceeding 15 30 seconds of error. Such results have been obtained with test drivers who have been asked to conduct in a flexible manner, by perceiving the information provided by the method without trying to "obey" it in a strict way, nor to react instinctively when this information evolves in time. [106] For information transmitted to a security operator to result in effective action on orders, the information must be accurate in terms of value, exploitable at the time it is perceived, collected and understood. correct way. To answer the problems prior to the invention described in point [001] in [022], an adapted HMI task (20) makes it possible to use the results of the preceding points to transmit to the driver all along the route, in real time and adapted the calculated information. At each of its cycles, the HMI task (20) performs the following steps: a) The HMI task (20) sets the start date of its cycle. b) For each Sequence Extraction Task in use, reliable on its last cycle, and used by the method to compute information: i) the HMI task (20) determines the period elapsed between the start date of said cycle of the HMI task (20) and the start date of said cycle of this Sequence Extraction Task, ii) the HMI task (20) adds to this period a period that compensates for the elapsed time between the [106.a] point and the point [106.bi] applied for said Sequence Extraction Task, iii) if the obtained period is greater than a threshold then the HMI (20) does not transmit the information concerned to the driver; otherwise the HMI (20) transmits this information to the driver. c) Of all the periods that the HMI task (20) calculated at [106.b] of this cycle, the HMI (20) determines the period which is the longest and stores it (period P1) so that the Sequence Extraction tasks use it on their subsequent cycles in point [029] calculations. In one embodiment of the present invention, the screen facing the driver is touch sensitive, which allows him to simply activate and deactivate the display of a flashing light-up callback. course. [107] Thus for each result to be transmitted to the driver, the HMI task (20) performs the following steps: determines the elapsed time since receipt of information from a source and the expected display time of said result, result determined by a task from said information; if said delay is below a threshold then transmits said result to the driver. [108] This method is particularly intended for driving trains. 35

Claims (13)

CLAIMS1) A method for assisting train drivers in their craft, said method characterized in that it comprises the following elements: Sequence Extraction Tasks (3x) which transmit information from sources (1x) to a Task d Analysis (4) which evaluates them and transmits a synthesis to the PKTRAIN task (6) which determines the curvilinear abscissa of the head of the train and transmits it to the following tasks related to the HMI (20): MIRLITONS (7) determines train-signal distance; EVEIL (9) alert of imminent braking; TOPFREINAGE (10) invites to brake and adapts the process; FOLLOWING (11) assists braking; VITESSEOPTI (12) determines the optimal speed. 2) A method according to claim 1 characterized in that at a frequency f each source provides the method information Longitude, Latitude, Cap, Speed, characterizing a fixed point relative to the front end of the train of a known distance by report at this end. 3) Process according to claim 1 characterized in that each of its cycles, each task performs the following steps: stores the start date of said cycle; if it uses information from another process task, then it compares the end date of the other task with its own cycle start date, if the result of this comparison exceeds a threshold then it declares itself out service and declares that said other task is out of service and declares that said information is unusable; memorizes the end date of said cycle; compare its own end-of-cycle date with its own start date, if the result of this comparison exceeds a threshold then it goes out of service. 4) Method according to claims 1, 2 and 3, characterized in that each Sequence Extraction Task of index x (3x) performs the following steps: extract from the data stream transmitted by the index source x (1x) longitude, latitude, heading, speed information; corrects the longitude and latitude information it has retrieved, corrections made according to the process execution time. 5) Method according to the preceding claims, characterized in that the Analysis Task (4) evaluates each Sequence Extraction Task according to the following steps: performs vector comparisons between the velocity vectors defined by said Sequence Extraction Task and velocity vectors defined by a plurality of Sequence Extraction Tasks; assigns a "reliable" characteristic or "unreliable" characteristic to said Sequence Extraction Task based on the results of these vector comparisons. 6) Method according to the preceding claims characterized in that the Analysis Task (4) calculates a centroid P and a velocity vector using the information transmitted by the "reliable" Sequence Extraction Tasks. 7) Method according to the preceding claims characterized in that the task PKTRAIN (6) performs the following steps: determines the point PK1 origin of an arc r of a curve C, and the point PK2 end of said arc r of said curve C , the head of the train being located on said curve C, said curve C and said arc f being defined by the plane line; determines the curve C passing through PK1 and PK2; determines the normal to the curve C, normal passing through P; determines the coordinates of the point P 'of the curve C closest to P; determines the curvilinear abscissa of P 'on the curve C; defines that the coordinates of the head of the train are those of P '; determines the abscissa curvilinear to the rail of the head of the train and deduces its PK. 8) Method according to the preceding claims characterized in that the task MIRLITONS (7) estimates the distance to the rail between the head of the train and signaling elements present on the route. 9) Method according to the preceding claims characterized in that the EVEIL task (9) performs the following steps: - estimates D1: the distance to the rail between the head of the train and the next scheduled stopping point; estimates the DINITIAL braking distance required for the train to stop; emits an alert if the difference between D1 and DINITIALE is less than a threshold. 10) Method according to the preceding claims characterized in that the TOPFREINAGE task (10) performs the following steps: issues an alert when is below a threshold the difference between Dl and DINITIALE; emits an alert when the distance D1 is below a threshold; stores each pair (Xi; Vi), i being an index incremented at each storage and reset at each braking, Xi being for a value of index i the distance to the rail separating the head of the train from the planned stopping point, and Vi being for a value of index i the speed of the train recorded at abscissa Xi; calculates the intercept a, the director coefficient b, and the correlation coefficient c, of the shape least squares curve: Vi = a + b - H (Xi); authorizes the task EVEIL (9) and the task TOPFREINAGE (10) to be used during the next braking the calculated results; compares the degree of braking performed with the degree of braking commanded and issues an alert if the result of this comparison exceeds a threshold. 11) Method according to the preceding claims characterized in that for a braking in progress the FOLLOWING task (11) estimates the correction to be made on the brake control. 12) Method according to the preceding claims characterized in that the VITESSEOPTI task (12) performs the following steps: for each fraction of the journey until the next scheduled stop, estimates the speed limit achievable technically and legally by the train; calculates a speed to make the schedule at the next scheduled stop; establishes whether said speed is relevant information according to regulatory and hourly limitations. 13) Method according to the preceding claims characterized in that for each result to be transmitted to the driver, the HMI task (20) performs the following steps: determines the time elapsed since the receipt of information from a source and the instant d expected display of said result, result determined by a task from said information; if said delay is below a threshold then transmits said result to the driver.