METHOD FOR DETERMINING AN OPTIMIZED ROUTE FOR AN ELECTRIC VEHICLE WITH REGARD TO TRAVEL TIME AND ENERGY COST

Abstract

The invention relates to a method carried out by a computing unit for determining an optimal journey (TRJ) of an electric vehicle on a predefined route, with regard to a weighting between a minimum travel time criterion and a minimum energy cost criterion, using reference data comprising road mapping information and information relating to the position and installed power of charging stations, a weighting coefficient (γ), embodying a preference between the minimum travel time criterion and the minimum energy cost criterion, defined by the user, the computing unit being configured to identify a minimum of an optimization function using a virtual cost of time spent on the predefined route and a virtual or real cost of the electrical energy expended, by a multi-domain dynamic programming method or by an interior point type non-linear programming method.

Description

Title of the invention: METHOD FOR DETERMINING A OPTIMIZED ROUTE OF AN ELECTRIC VEHICLE WITH REGARD TO TRAVEL TIME AND COST ENERGY

[0001] The invention relates to a method for determining an optimal route for an electric vehicle with regard to the journey time and the energy cost, and in particular based on a preference weighting provided by the user of the vehicle.

[0002] The optimal route is thus personalized according to the user's wishes and preferences.

[0003] The focus here is on all-electric vehicles, i.e. vehicles without an internal combustion engine. In addition, the focus here is on journeys of considerable length, for example at least 400 km, or more, which generally requires one or more stops to recharge the vehicle's battery. In a typical case, the majority of the planned journey is to be made on the motorway network, equipped with a plurality of charging stations.

[0004] The method may include the actual completion of the route according to the indications and recommendations resulting from the calculation of the optimal route.

[0005] Regarding the amount of energy stored in the battery, in practice we are talking about several tens of kWh. A 100% electric vehicle battery has an energy storage capacity typically between 40 kWh and 100 kWh, without these values ​​being limiting. To set the order of magnitude, a conventional or utility type electric vehicle consumes between 15 kWh and 25 kWh per 100 kilometers traveled.

[0006] In this type of vehicle, it is necessary to recharge the battery. Recharging can be carried out from an electrical outlet available in the usual home of the vehicle user (so-called 'domestic' outlet) or from a charging station located in the public or semi-public domain.

[0007] For the journeys of interest considered here, except where applicable at the departure and arrival points, the vehicle uses the charging stations available in the public domain along the journey to be made.

[0008] Generally, a reference database is available comprising road mapping information and information relating to the position and installed power of charging stations. The mapping data includes static data (length and gradient on each section of the journey, altitude, etc.) and non-static data (for example the average speeds observed on each section of the journey, the usual wind or the wind of the day, the weather forecast for the day of the journey, in particular the outside temperature, etc.).

[0009] Concerning the vehicle of interest, we also have parameters relating to the vehicle, including in particular at least its mass and a coefficient of penetration in the air which characterizes the aerodynamic drag forces which can prove to be significant or even preponderant at cruising speeds above 110 km / h.

[0010] As will be seen later, the parameters relating to the vehicle also include parameters relating to the battery, to its thermal behavior, knowing that it is necessary to cool it effectively during a high-power recharge known as rapid recharge.

[0011] In view of the above, it is intuitively understood that the energy consumption required to complete the journey depends not only on cruising speeds but also on the need to cool the battery during the execution of a fast charge. Also, it is possible to limit the cruising speed and opt for slower charges, but this will be to the detriment of the journey time.

[0012] For a long journey with multiple recharging possibilities, we understand that it is impossible to achieve the optimization desired by the user via a simple calculation and even using a basic calculation tool.

[0013] Only a high-performance, well-documented computing unit, embedded or partly using a remote service, can effectively identify the best solution, in a response time of a few seconds, compatible with what is expected of a planning agent which may have to replan the journey at any time, or which may have to react to the unavailability of a charging station.

[0014] To this end, the present invention proposes a method for determining an optimal journey of an electric vehicle on a predefined route between a starting point and an arrival point, with regard to a weighting between a minimum journey time criterion and a minimum energy cost criterion, the method comprising: EA- have a reference database including road mapping information and information relating to the geo-position and installed power of charging stations,

[0015] EB- have a behavioral model of the vehicle of the electric vehicle, comprising a plurality of parameters in particular at least one mass and one coefficient of penetration in the air, EC - choose a weighting coefficient y, embodying a preference between the minimum travel time criterion and the minimum energy cost criterion, y being between 0 and 1, and optionally a desired state of charge at the arrival point, ED- define a virtual cost function noted J of which a minimum corresponds to an optimal journey, / / J-Lo( (ir)^(O + yLe(t) ) d / Where Lt represents a virtual cost of time spent and Le represents a virtual cost of electrical energy spent, to is the starting time and tf is the arrival time of the journey, EF- identify, using at least one calculation unit (4), at least one minimum of the virtual cost function J, df, . , , „ . . formula in which the , min J = J, o ((ly)Z.,(O + yL,P)) df ]vvJ'cty, [sSOCjfSOC f} current speed vv and charging power Pch are calculation input variables, in which the current position of the vehicle noted s, the current state of charge noted SOC are states, L represents the known departure time and tf represents the arrival time unknown a priori, SOCf is information on the desired charge state at the arrival point, EG- deduce the charging stations where the vehicle must stop to recharge and with which charging parameters, and the set driving speeds on each section of the journey,

[0016] Thanks to the provisions promoted above, it is possible to identify the optimal route with the target speeds and the most relevant recharging stops with regard to the weighting coefficient y chosen, y being chosen by the user.

[0017] For example, when the user chooses y=0, he favors the shortest travel time, whereas conversely, when the user chooses y=1, he favors the lowest energy expenditure.

[0018] It should be noted that the method then provides for actually carrying out the journey according to the indications and recommendations resulting from the previous calculation, which implements a concrete and practical application of the previous calculation.

[0019] It should be understood here that the term “state of charge” should be understood here in the broad sense, namely it can designate the state of charge itself (“SOC”) or the state of energy (“SOE”) of the battery depending on the calculation method used.

[0020] It should be understood here that the information "desired state of charge at the arrival point" can be narrow information such as a target to be reached, or more flexible information such as an interval of states of charge in which the battery must be at the arrival point of the journey, or even an even more flexible expression with a single inequality, for example the user wishes the state of charge at the arrival point to be greater than or equal to a minimum state of charge desired at the arrival point of the journey.

[0021] It should be understood concerning step EG-, that the “charging parameters” include in particular the target recharging current, that is to say the target recharging power, in particular the target power at the start of the charging sequence, this power being able to decrease with the progression of the charging according to the electro- battery chemistry and its temperature behavior.

[0022] As will be seen later, it should be noted that the quantity of “virtual cost of electrical energy spent” can be valued economically or not.

[0023] Preferably, the behavioral model of the vehicle of the electric vehicle comprises a dynamic behavioral model generally comprising coefficients of resistance to the advance of the vehicle including in particular but not exclusively the coefficient of penetration in the air.

[0024] According to one embodiment, the behavioral model of the vehicle further comprises a behavioral model of the battery, in particular from a thermal point of view.

[0025] Thus, the calculations made for the function J take into account, on the one hand, during the charging phases, heating of the battery as a function of the recharging current, and on the other hand during the driving phases, as a function of the maximum tolerable temperature in the battery, with, where appropriate, a reduction in the power delivered.

[0026] According to one embodiment, the virtual cost of electrical energy expended (Le) is expressed in kWh, i.e. in energy consumption not economically valued.

[0027] As a result, the user can focus on minimizing the carbon footprint of his journey, without taking into account the cost of acquiring the necessary electrical energy.

[0028] According to one embodiment, the virtual cost of electrical energy expended (Le) is expressed in monetary values, the energy consumption being economically valued as a function of the respective purchase costs of the electrical energy from each charging station requested for the completion of the optimal journey.

[0029] Advantageously, the energy criterion is then expressed as the monetary value of the expenditure to be made in electrical energy on the journey. In this case, the user prefers the financial criterion to the carbon footprint criterion.

[0030] According to one embodiment, the minimum of the virtual cost function is obtained by a multi-domain dynamic programming method.

[0031] Advantageously, the exploration domain of the calculation depends on the phase of the journey, namely whether a recharging sequence is in progress or conversely whether no recharging is in progress and the vehicle is in a general driving situation.

[0032] According to one embodiment, there is thus a reduction of one degree of freedom. The recharging current is zero in the driving situation while the geographical position is fixed (fixed distance variable) when the vehicle is in the charging situation.

[0033] According to one embodiment, the calculation of the function J presents a distinction between a charging sequence situation where the battery behavioral model applies to the first order and a driving situation where the dynamic behavioral model of the vehicle applies to the first order.

[0034] According to one embodiment, the minimum of the virtual cost function is obtained by a interior point type nonlinear programming method.

[0035] According to one embodiment, the calculation introduces at least one additional variable and uses a so-called Lagrangian formulation to identify the values ​​of variables corresponding to the minima of the function J.

[0036] According to one embodiment, the dynamic programming method uses boundary lines and / or surfaces.

[0037] According to one embodiment, the predefined route has a length of at least 500 km and requires at least two electric charging stops.

[0038] The present invention also relates to a system comprising at least one calculation unit configured to implement the method as described previously.

[0039] The present invention also relates to a vehicle comprising at least one computing unit configured to implement the method as described previously.

[0040] The invention will be further detailed by the description of non-limiting embodiments, and on the basis of the appended figures illustrating variants of the invention, in which: [Fig.l] is a schematic representation of the profile of a motor vehicle; [Fig.2] shows an exemplary block diagram of the traction / propulsion system and user interface implemented in the present invention; [Fig.3] shows a diagram illustrating the possible results depending on the weighting coefficient chosen by the user; [Fig.4] schematically represents a vehicle route from a starting point so to an arrival point sf, with charging stations CS; along the route; [Fig.5] shows an example of a functional diagram of the method implemented in the present invention; [Fig.6] schematically represents a vehicle journey from a starting point so to an arrival point sf, with different solutions identified depending on the value of the weighting coefficient chosen by the user.

[0041] In the various figures, the same references designate identical or similar elements.

[0042] General information

[0043] In [Fig.l], a VH vehicle is shown schematically. The vehicle in question may be a passenger vehicle, a utility vehicle, a van, a recreational vehicle, a minibus, a coach, etc.

[0044] The vehicle in question VH comprises an electric traction chain with an electromotive group which comprises an electric machine 1, for traction or propulsion, depending on whether the front axle or the rear axle is concerned.

[0045] As seen in [Fig.2], the electrical machine 1 is controlled by a power module 2 also called in practice 'inverter' (noted in short INV). A calculator control 3 controls the electric machine via 1 the power module 2. Depending on the circumstances of use, the electric machine 1 can be controlled as a motor or controlled as a generator as known per se, in the case of regenerative deceleration or braking.

[0046] The vehicle VH is a 100% electric vehicle. The powertrain comprises a traction battery 5. The traction battery 5 has an energy storage capacity typically between 40 kWh and 100 kWh, without these values ​​being limiting.

[0047] In the context of the present invention, the battery 5 of the vehicle is equipped with a battery management computer 15, otherwise known in the trade as BMS (Battery Management System).

[0048] The general architecture and functions of a BMS calculator are known per se, this unit can determine the amount of energy entering the battery by means of one or more current sensors and one or more voltage measurements at the terminals. This BMS unit monitors the internal temperature of the battery and can decide to limit the incoming or outgoing current in the event of excessively high temperature.

[0049] The vehicle VH comprises an on-board charger marked 12, otherwise known in the trade as OBC from the English 'On Board Charger' connected to the charging socket 11 accessible from outside the vehicle. The general architecture and functions of the on-board charger are known per se, this unit ensures electrical protection with respect to various possible external connections. This unit ensures in particular the AC / DC conversion (alternating to direct) in the case of charging from an alternating current source.

[0050] Furthermore, fast charging is a need during long journeys where the distance traveled is greater (or even much greater) than the range allowed by the battery capacity. There is currently an increase in the number of fast charging stations with terminals capable of delivering powers of 50kW, 75kW, 150kW, 300kW, 350kW, without excluding other intermediate calibers. It should be noted that these fast recharges use a transmission of electrical energy in the form of direct current and require the use of a specific charging cable having large cross-section conductors. The current supplied by the terminal enters directly into the battery without passing through a transformer, i.e. without AC / DC conversion (alternating to direct), with better energy transfer efficiency.

[0051] Depending on the power transferred, heating of the battery 5 can be quite significant and it is necessary to evacuate the calories by running a pump and a fan which consume electricity, and therefore degrade the energy balance of the recharge.

[0052] The battery 5 is equipped with at least one temperature sensor 51 connected to the BMS 15.

[0053] The vehicle comprises a multifunction display 6, including in particular, as known per se, a road navigation function.

[0054] The vehicle may further comprise a supervisory computer 4 whose functions will be discussed later.

[0055] The supervisor computer 4, the control computer 3 of the electric machine, the battery management computer 15, the multifunction display 6, and the on-board charger 12 are entities called here on-board computers because they are located in the vehicle, they are connected to each other by a multiplexed bus 44 in a manner known per se. In contrast, certain other computer and electronic means are not located on board the vehicle, in particular in the context of the present invention, this is the terminal controller placed in the charging terminal.

[0056] Furthermore, one (or more) application server(s) 18 may be provided, and a smartphone 19 otherwise called here a communicating multifunction portable terminal, which may be involved in the calculations which are set out below. The vehicle equipment communicates with these external means via one or more wireless links 45.

[0057] It is noted that the vehicle is usually equipped with an air conditioning function. When the outside temperature is high, the air conditioning function is activated and its operation requires electrical energy consumption.

[0058] We are interested here in TRJ journeys of significant length, for example at least 400 km, or more, which generally requires one or more stops to recharge the vehicle battery.

[0059] In the example illustrated in [Fig.6], the journey has a length of 1200 km. This is the case, for example, of a journey from the city of Brest to the city of Avignon.

[0060] In a typical case, the majority of the planned journey is to be made on the motorway network, equipped with a plurality of charging stations. In [Fig.4], the first seven charging stations on the route are shown, namely CSi, CS2, CS3, CS4, CS5, CS6, CS7. The total number of stations is noted ch. Each charging station comprises a certain number of independent individual terminals.

[0061] The optimal route TRJ of the vehicle is formed as a predefined route between a starting point so and an arrival point sf. In [Fig.4], the first five end points of sections sb s2, s3, s4 have been identified. The division into sections allows for a certain homogeneity of parameters on each section.

[0062] Functional need and associated method

[0063] The driver of the vehicle, or more generally a user, can choose to give priority to the shortest possible journey time or to give priority to the lowest possible electrical energy consumption, or advantageously in the case of the present invention he can choose a parameter balancing between the two criteria.

[0064] Thus the user chooses a weighting coefficient y, embodying a preference between the criterion of minimum travel time and the criterion of minimum energy cost.

[0065] y being between 0 and 1. When the user chooses y=0, he favors the shortest travel time, whereas conversely when the user chooses y=1, he favors the lowest energy expenditure. When the user chooses for example y=0.40, he wants a compromise between the shortest travel time and the lowest energy expenditure.

[0066] [Fig.3] illustrates an example of results obtained using the method promoted here, by means of 3 points 81, 82, 83 respectively representing three distinct solutions.

[0067] Point 81 illustrates the result of the optimization calculation when the user has entered a small weighting coefficient y, the preferred criterion then being a short travel time. Conversely, point 83 illustrates the result of the optimization calculation when the user has entered a weighting coefficient y close to 1, the preferred criterion then being low energy consumption (or a low purchase price of the electrical energy consumed). Point 82 illustrates the result of the optimization calculation when the user has entered a median weighting coefficient y close to 0.5.

[0068] The vertical line TTmin represents the theoretical minimum travel time when the speed practiced is equal to the maximum regulatory speed, respectively on each section.

[0069] In addition, the user optionally chooses a desired state of charge at the arrival point noted SOCf.

[0070] The information about the desired state of charge at the arrival point SOCf is information given by the user. The information about SOCf can be narrow information such as a target to be reached, for example 50%, or more flexible information such as an interval of states of charge in which the battery must be at the arrival point of the journey, for example between 30% and 60%, or even an even more flexible expression with a single inequality, for example the user wants the state of charge at the arrival point to be greater than or equal to a minimum state of charge desired at the arrival point of the journey, for example > 25%.

[0071] The vehicle has at its disposal a reference database comprising on the one hand road mapping information, as known per se in navigation systems, and on the other hand, information relating to the DB-CS charging stations. The information relating to the charging stations (CSch) includes for each charging station the geo-position and the collective installed power, and the individual installed power for each charging terminal of the station. The general availability of each terminal can also be included, as can the schedule of the day reservations if applicable.

[0072] The road mapping information includes for each section static data (length and gradient on each section of the journey, altitude, etc.) noted Map-St in [Fig.5], and non-static data, noted Map-nSt in [Fig.5]. The non-static Map-nSt data may be, for example, the average speeds observed on each section of the journey, the usual wind or the wind of the day, the weather forecast for the day of the journey, in particular the outside temperature.

[0073] We also have a behavioral model of the MDV vehicle. The behavioral model of the vehicle includes in particular at least its current mass (rolling mass of the moment) and an air penetration coefficient which characterizes the aerodynamic drag forces. As will be seen later, there are also other coefficients representing the dynamic behavior of the vehicle.

[0074] The vehicle behavioral model is based on the following equations.

[0075] The instantaneous electrical power balance during rolling would be as follows.

[0076] Pb(t)=ViP(l)+PBnI(t)+Pa,a dr

[0077] Pdr ( t ) represents the electrical power involved in the mechanical displacement of the vehicle,

[0078] Pbtm(t) is the power consumed by the thermal management of the battery, expressed in Watts.

[0079] P aux represents the electrical consumption of the auxiliary equipment (by example the air conditioning compressor, headlights, windshield wipers, interior electrical equipment, etc.) expressed in Watts.

[0080] represents an efficiency parameter between the electric machine and the battery with l = sign (P^),

[0081] = +fi2Fd(t

[0082] wi t) is the feed rate in [m / s]. F^i) is the traction force, expressed in newtons [N].

[0083] Pj, and are positive calibratable parameters and are coefficients relating to a map of electrical consumption of the electrical machine. More precisely, expressed in [W / m / s] represents the friction losses, P-, represents the main electromechanical term and / 3^ expressed in [W / NA2] represents the Joule effect losses.

[0084] More precisely, the displacement of the vehicle is defined by the following longitudinal dynamic equations.

[0085] yj where s is the position coordinate along the route, vv is the speed in advance.

[0086] = J- (nrlFjt) - a2 (v (t) - wH. (t))2 - a^v (?) - a0 - g mvsïn (3)) mv is the mass of the vehicle [kg], loaded and towing a trailer if necessary, is the transmission efficiency, dimensionless is the wind speed in [m / s] a2 is the aerodynamic drag coefficient [N / (m2 / s2)] ai is the rolling resistance coefficient [N / (m / s)] is the friction coefficient [N] S represents gravity [m / s2] 0 is the road gradient [deg]

[0087] al, a2 and a3 are calibratable parameters.

[0088] The tensile force Fd(t) can be calculated as follows.

[0089] = mv + a2(vv(t) -w(r) + )

[0090] Where the vehicle is assumed to maintain a constant speed vv on each section, so namely dvv / dt close to 0.

[0091] It should also be noted that the regenerative traction or deceleration force is limited by the following formulas.

[0092] max (Preg / im, Pdr^in (v)) < Pdr (v) < Pdrmax (v)

[0093] Fdfnin(y) ï Fd(v) < F^J^y)

[0094] Where Pregj if a, cl in and Pd>-.ma.\ are respectively the regeneration limits and the minimum and maximum powers of the electric machine. Fdjnit, and Fd^wx are respectively the force limits of the electric machine.

[0095] The behavioral model of the vehicle also includes a behavioral model of the battery cooled by a cold plate.

[0096] An example of a behavioral model of the battery is given below.

[0097] Vc(') = r„(0-R(SOC, T„ sign(i() 100981 7F = +Pc,.,l(.t')+P>,„M+Pmb) is the cell voltage in [V] R is the internal resistance of the battery in [Ohms] Tb is the battery temperature [°C] mc is the mass of the battery cell [g] This is the specific heat of the battery cell [J / K / kg] Pgen is the power generated by heating the battery [W] Pcool P beat are the cooling and heating powers of the system battery thermal control t [W] Pamh is the heat exchange power with the environment [W]

[0099] In particular we can write:

[0100] P (t\JHTCxS ifT^T^ t>wl 0, otherwise

[0101] P (ti = lHTCxS( ' 'f Tb < T-> ' heat X1 / ni ■ 0, otherwise [01021) = HTCa xS(Tb(t)- Tmb(t))

[0103] where TC(m[ and Tamb are respectively the temperatures of the cooling circuit and the ambient, HTC and HTCa are respectively the transfer coefficients with respect to the cold plate and the environment, each expressed in [W / K / m'2], where S in [m] is the contact surface of the battery cell with respect to the cold plate.

[0104] The electrical power Pp-py(t) involved by the thermal management of the battery can be written as follows

[0105]

[0106] Ppump + (tamb)PcaoM, siPcool>Q Ppump ~ hhPheat (t), If Pheat < 0 otherwise Ppump is the power consumed by the pump of the thermal management system of the battery.

[0107] In addition to these models, the journey can be divided between driving phases and charging phases, and the behavior of the battery will follow the following logic according to said phases Pb ( t ) — Vb ( t ) i^t) in each phase.

[0108] With these elements, the drum dynamics are written as follows.

[0109] = A T T b (t)+B T 2 1 / ' + C T T aml lt

[0110] [YES] Identification of the optimal route

[0112] We define a virtual cost function noted J, expressed as follows. J^JU-ïOMO +?M0) dz

[0113] A minimum of J corresponds to an optimal path.

[0114] Lt represents a virtual cost of time spent and Le represents a virtual cost of electrical energy spent, to is the starting time and tf is the arrival time of the journey.

[0115] We identify, using at least one calculation unit 7 (supervisor calculator 4 or other calculation unit), at least one minimum of the virtual cost function J, expressed mathematically as follows. min J= L ( ( lr)M0 +r MO ) & [ vv,Pch}, faSOCjpSOCf |

[0116] The current speed vv and the charging power Pch are calculation input variables. The current position of the vehicle noted s, the current state of charge noted SOC( / SOE) are states.

[0117] to represents the known departure time and tf represents the arrival time unknown a priori, SOCf is information on the desired load state at the arrival point.

[0118] Two distinct methods are proposed to identify one or more minima of the function J.

[0119] By identifying one or more minima, it is possible to deduce the charging stations CSai where the vehicle must stop to recharge and with which charging parameters, and the set driving speeds Wj on each section of the journey, as shown in [Fig.5].

[0120] Logically, the user then actually makes the planned journey according to the indications and recommendations resulting from the previous calculation, with preferential display on the multifunction screen 6 of the vehicle, or even on the smartphone 18.

[0121] According to one option, the virtual cost of electrical energy spent Le is expressed in kWh, i.e. in energy consumption not economically valued.

[0122] According to another option, with financialization, the virtual cost of electrical energy spent Le is expressed in monetary values, for example in euros, the energy consumption being economically valued according to the respective purchase costs of electrical energy from each charging station requested for the completion of the optimal journey.

[0123] On this subject, we generally note that the price of the kilowatt hour is proportional to the installed power of the charging station. Therefore, depending on whether we want to minimize the cost of the journey, it is better to recharge at stations with low or medium power, but at the expense of efficient journey time.

[0124] For the rolling phases the variables are expressed according to the distance domain along the curvilinear abscissa.

[0125] Conversely, for the recharging phases, we use variables expressed in the domain of the state of charge SOC.

[0126] These transformations of variables give the following expressions

[0127] min 7 = 1^1^((1-^0)+7^^)+^^^ 07^7((^^^(^)^^(5^))!

[0128] It should be noted that the term ^^5^niax{0 ( T} ( v) - Ti )} 2 is added to represent the degradation of the battery which remains at high temperature.

[0129] However, this function can be represented by any function that represents battery aging.

[0130] If s(t) corresponds to a position of charging stations:

[0131] SOC^SOC^s^ y T^SOC^-T^s-)

[0132] / F ; [ SOCC ^ = ^^\ A ^ S0 C+B rf ' + C t T„„(SOC)

[0133] Pch(SOC) = Pb(SOC)-PBTm{SOC}-Paux

[0134] Ihmin(SOC, Tb) < ib(SOC) < O

[0135] _ Pcjvnax ù P cb(SOC)

[0136] If s(t) does not correspond to a position of charging stations:

[0137] SOC(sH)=SOC>, Tb(sU) = Tb(SOCf), fori>\

[0138] dSOC _ __1 <ts vG j PswÀc ( ‘0

[0139] £±=_1_ fl S" pf A TfÂs) +Bt ï‘c-W2 + C7 Tmb(s) j

[0140]

[0141] PM=P p.<M =z dr(S) +PbTm(S 1^(5) +P2Fd(s JcooM. +Paux )v(5) +f CSC J d

[0142]

[0143] h^isoc, Tb) S z;.(s) < Ibwx(SOC, T b)

[0144]

[0145] Pdjnin ( ) m-^PregJi iFd(s) <Fd / nax(v') n,P .{v}\<Pdr{s} <p drjmnx z drjna^f)

[0146] tout en satisfaisant les conditions suivantes :

[0147] v.m="(SOC)" -r(soc, tb, sign(4) ) ic(t)

[0148]

[0149]

[0150]

[0151] pi,(r)="v,M" =1's v (ï)>iM P^J) =V,flTCxS

[0152]

[0153]

[0154]

[0155] S0C,nln(t); Pheaunax (7) PsT / O ~ iSOC(t) <SOCmax(t) \t)<Vfncix(t) iPdis(t) <Pmol„aÂt) Ppump + t]c ( Tamb ) Pdis ( t ), i f Tb >Tcooi Ppump ~ fPdis ( 0 ' b < Pcool

[0156] s(h) = ¾ la SOCM^SOCq, Tb ifTh^TCO(A (fo) ~ Pbfi

[0157] s(tf) = c SOC^tf) ex 7- C tf) — £Xfh

[0158] Discretization

[0159] We set sk = So + $k, ÆeK = {0,1, ..., K] for the rolling phases.

[0160] We set sOCq = SOCl0 + q, q GQ = {0,1, ..., For the Phases of charSe-

[0161] We can also write = and _ soc^-soc^ K Tî~ Q

[0162] Multi-domain Dynamic Programming

[0163] In this method, inputs and states are quantized in mesh grids covering the range of achievable values ​​for each variable. This allows many possible combinations of state transitions to be evaluated depending on the input values ​​used.

[0164] A dynamic programming algorithm is composed of two steps: a backward loop where calculations are performed to find the optimal control strategies, and a forward loop where the resulting strategies are applied to the system based on the initial state values. The main advantage of the proposed multi-domain algorithm is the adaptation of these steps to "reduce" one of the dimensions of the problem when moving from the driving phase to the charging phase.

[0165] In classical dynamic programming, the backward loop is executed backward along the independent variable (time, distance, SOC, etc.), starting from the end of the horizon to the beginning. Considering x as the state, u as the input and k as the index of the independent variable. [°166] J^x^ =min {g(xk,uk) +Jt+1(xk+l)}, V k = {Kl, ...,1,0} uk

[0167] where J^(xk) is the marginal cost function at point k with = dx-x) being a final cost and #(¾ uk) the instantaneous cost.

[0168] Here, the calculations are performed for all state and input values ​​in the mesh grids. However, the proposed methodology requires moving from one independent variable to another, which is part of the state grid, for example, being in the distance domain with SOC and Tb as states and moving to the SOC domain while keeping Tb as the state of mesh grids, each variable that is grouped, for example, in the travel management problem, an upstream main loop is defined in the distance domain while an inner SOC loop is added when reaching a charging station.

[0169] In these cases, the upstream and downstream loops are modified by creating nested loops for each variable that is compressed. For example, in the travel management problem, a main upstream loop is defined based on distance, while an inner SOC loop is added when a charging station is reached.

[0170] A duplication of the locations of the charging stations is added to the distance horizon to take into account the steps preceding and following a possible recharge.

[0171] In order to preserve the numerical precision of the strategy, the use of boundary lines (also called limit lines) is incorporated into the algorithm.

[0172] The construction of the boundary lines starts from the edges of the terminal set, where the transition from step kN to kN - 1 is defined with the inputs that generate the maximum and minimum variations of the SOE according to the dynamics of the inverted battery. Then, if a constraint is violated, the input at the boundary of this constraint is selected and the corresponding state is changed.

[0173] This procedure is repeated in reverse until k = 0.

[0174] Additionally, the lower boundary line is reset to SOEmin at each station charging to take into account possible recharges.

[0175] Furthermore, the performance of the algorithm is further improved by the use of variable mesh grids for the discretization of the state and input spaces.

[0176] Using these grids, the mesh values ​​are adapted according to the input and SOE limits given by the constraints and boundary lines at the input space at each distance step.

[0177] Nonlinear programming

[0178] Alternatively, the problem can be solved using nonlinear programming techniques to improve the computational performance of the solution. It is proposed to use an interior point method here.

[0179] In an interior point method, inequality constraints are transformed into equality constraints by adding soft variables, which are handled in the problem at using a barrier or penalty function.

[0180] As an example, consider the following optimization problem

[0181] min a st A(u) = Q g(u)> 0 inin f(x) X st c(x} = 0 y >0 Wh x = [h , y] min. f (x) - g Y hv yj •X

[0182] The Lagrangian is defined as follows:

[0183] L = / ( x ) - / z £ , = Jn ( y. ) + 2

[0184] Then, a solution to the problem is obtained by satisfying the Karush-Kuhn-Tucker™ (KKT) conditions using a linear search method.

[0185] There are several interior point solvers on the market, such as the “fmincon™” function of Matlab™ or other free solvers such as “IPOPT™”. In the present invention, the so-called “IPOPT” solver is preferred due to its computational performance, but similar results can be obtained with other solvers.

[0186] Turning to [Fig.6], the upper part of the graph represents the state of SOC charge as a function of distance (abscissa s) for three distinct solutions resulting from calculations with different parameters, notably the weighting y, but for an identical route. The lower part of the graph represents the battery temperature profile for one of the identified routes. The middle part of the graph represents the forward speed vv of the vehicle along the route for one of the routes traveled.

[0187] Curve 91 represents a fairly fast journey with 4 stops to recharge the vehicle's battery. Curve 92 represents a slower journey at cruising speed but with only 2 (longer) stops to recharge the vehicle's battery. Curve 93 represents a somewhat unusual journey where the vehicle stops to recharge at each of the charging stations available along the route.

[0188] Between the respective initial and final times t0, tf, the times tri, tr2, tr3, tr4, materialize the start times of each recharging sequence. It is noted that, for the purposes of modeling, particularly in distance, the position which represents the recharging station is duplicated, namely a point corresponding to the arrival state of the vehicle and another point corresponding to the re-departure of the vehicle.

[0189] We note that the battery temperature, illustrated by the curve at the bottom corresponding to the first fairly rapid journey (curve 91) shows temperature peaks at each charging location. ​

Claims

Claims

1. Method for determining an optimal journey (TRJ) of an electric vehicle on a predefined route between a starting point (so) and an arrival point (sf), with regard to a weighting between a minimum journey time criterion and a minimum energy cost criterion, the method comprising: EA- have a reference database comprising road mapping information and information relating to the geo-position and installed power of charging stations, EB- have a vehicle behavioural model (VBM) of the electric vehicle, comprising a plurality of parameters, in particular at least one mass and one air penetration coefficient, EC - choose a weighting coefficient y, embodying a preference between the minimum travel time criterion and the minimum energy cost criterion, y being between 0 and 1, and optionally a desired state of charge at the arrival point, ED- define a virtual cost function noted J, a minimum of which corresponds to an optimal path, Where Lt represents a virtual cost of time spent and Le represents a virtual cost of electrical energy spent, to is the starting time and tf is the arrival time of the journey, EF- identify, using at least one calculation unit (7), at least one minimum of the virtual cost function J, j vv,Pch], {sWC,tfiSOCf} formula in which the current speed vv and the charging power Pch are calculation input variables, in which the current position of the vehicle noted s, the current state of charge noted SOC are states, E represents the known departure time and tf represents the arrival time unknown a priori, SOCf is information on the desired state of charge at the arrival point, EG- deduce the charging stations where the vehicle must stop to recharge and with what charging parameters, and the set driving speeds on each section of the journey.

2. Method according to claim 1, characterized in that the behavioral model of the vehicle (MDV) further comprises a behavioral model of the battery, in particular from a thermal point of view.

3. Method according to any one of claims 1 to 2, characterized in that the virtual cost of electrical energy expended (Le) is expressed in kWh, i.e. in energy consumption not economically valued.

4. Method according to any one of claims 1 to 2, characterized in that the virtual cost of electrical energy expended (Le) is expressed in monetary values, the energy consumption being economically valued as a function of the respective purchase costs of electrical energy from each charging station requested for the completion of the optimal journey.

5. Method according to any one of claims 1 to 4, characterized in that the minimum of the virtual cost function is obtained by a multi-domain dynamic programming method.

6. Method according to any one of claims 1 to 5, characterized in that the calculation of the function J presents a distinction between a charging sequence situation where the battery behavioral model applies to the first order and a driving situation where the dynamic behavioral model of the vehicle applies to the first order.

7. Method according to any one of claims 1 to 4, characterized in that the minimum of the virtual cost function is obtained by a non-linear programming method of interior point type.

8. Method according to claim 7, characterized in that the non-linear programming method uses boundary lines and / or surfaces.

9. System comprising at least one calculation unit (4,19,18), configured to implement the method according to one of claims 1 to 8.

10. Vehicle comprising at least one computing unit (4,19,18) configured to implement the method according to one of claims 1 to 8.

Images