ACTIVE TUNING SYSTEM FOR ENGINE CONTROL UNIT USING AIRFLOW METER TABLE

Abstract

A method for an engine control unit to conduct real time updating of data tables based on feedback from the air flow sensor without external computing. The air fuel ratio is based on information received from an air flow meter, where a look-up table is actively tuned for the air flow calibration instead of the volumetric efficiency tables. Here, the look-up table involves a single dimension value so the calculations are simpler than more complex evaluations previously set forth.

Claims

1. A method for updating an automotive fuel injection system of a combustion engine during active tuning, comprising: reading an air flow meter for determining a measured air flow entering the combustion engine; recalling a target lambda (air-to-fuel ratio) value using the lambda value from one of a look-up table and an engine control unit calculated value; determining if a difference between a current lambda value and the target lambda value is present; adjusting the look-up table lambda adjustment value based on the difference between the current lambda value and the target lambda value; replacing the lambda adjustment value for the air flow in the look-up table with the adjusted lambda value; and wherein the air to fuel table values are checked when updating to ensure that successive values are larger than the preceding values.

2. (canceled)

Description

BRIEF DESCRIPTION OF THE DRAWINGS

[0020] FIG. 1 is a schematic of an automobile engine control system;

[0021] FIG. 2 is a schematic showing the flash and RAM memory;

[0022] FIG. 3 is a flow chart showing the start-up method for the ECU;

[0023] FIG. 4 is a flow chart for adjusting the active tuning using a knock count;

[0024] FIG. 5 is a flow chart for adjusting the knock count adjustment value;

[0025] FIG. 6 is a computer interface entry table for adjusting the active tuning;

[0026] FIG. 7 is a graph showing the delay in the lambda in response to a change in engine conditions;

[0027] FIG. 8 is a graph showing the delay in lambda due to the sensor being downstream of the engine;

[0028] FIG. 9 is a flow chart for setting the engine conditions using the active tuning;

[0029] FIG. 10 is a flow chart for checking the engine condition parameters are within limits;

[0030] FIG. 11 is a flow chart for adjusting the target lambda adjustment value;

[0031] FIG. 12 is a flow chart for calculating the midpoint of the index values; and

[0032] FIG. 13 is a flow chart for an alternate method of adjusting the fuel air requirements.

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS

[0033] FIG. 1 illustrates the components of an engine control system for a vehicle of the kind that can benefit from the present invention. The system begins with a power supply, such as the battery 5. The battery is connected to an ignition switch 10 that controls the power to the engine control unit 15, or ECU. A main relay 20 receives an input signal 25 from the ignition switch 10, so as to operate the fuel pump, and supply power to the fuel injectors. In addition to the power via the ignition switch 10 and the signal 30 from the main relay 20, the ECU 15 also receives a signal 45 from an oxygen sensor 35 on the exhaust system 40, which provides feedback on the conditions of the fuel burning efficiency. The engine's throttle position is conveyed to the ECU 15 by a signal 55 from the engine's throttle 50. In preferred embodiments, the ECU also receives a signal 60 from an engine knock sensor 65 as well as a signal 70 of the engine's crank position via the crank position sensor 98. The output of the ECU is a signal 75 for instructions to the fuel injectors 80 on the engine 85, and the firing instructions 90 on the engine's spark plug 95. The system operates as a feedback loop, where each input controls the output and vice versa as the ECU attempts to maintain an optimal fuel burning system despite changing conditions inside and outside of the system.

[0034] The engine 85 may also be equipped with an air flow meter 92 used to find out the mass flow rate of air entering the fuel-injected engine. The air mass information is necessary for the ECU 15 to balance and deliver the correct fuel mass to the engine. Air changes its density as it expands and contracts with temperature and pressure. In automotive applications, air density varies with the ambient temperature, altitude, and the use of forced induction, which means that mass flow sensors can be more appropriate than volumetric flow sensors for determining the quantity of intake air in each cylinder.

[0035] To aid in the foregoing performance of the ECU, after market engine control units often include tables that are updatable, but this feature is not found on OEM ECUs. The OEM ECU in a vehicle uses fixed tables in flash memory to store engine operational data. These tables may be directly referenced by flash memory program code, or they can be referenced by a pointer to the table. Because flash memory takes considerable time to erase and re-flash, it cannot be altered while the ECU processor is running as the ECU operation will halt while the re-flashing task is underway. Random access memory, or RAM, on the other hand, is fast to update but is un-initialized when the power is removed from the ECU and therefore is cleared every driving cycle.

[0036] To enable rapid real time updating of the ECU's tables, the data may be copied from flash memory to RAM. To achieve this, an area of ECU RAM is reserved for a copy of the table. At startup, the ECU's code copies the data from its flash memory to RAM. All program code pointers to the table are updated to reference the table in RAM. In addition, a communication routine is added to the ECU that alters the RAM table based on commands received from external tuning software on a laptop. When the ECU is powered down and re-started, the ECU startup routine copies the original flash memory table into RAM so that all changes made to the RAM table are lost. To prevent this, the laptop can detect that the ECU has re-started and update the table automatically.

TABLE-US-00001 Pseudo code - ECU RAM initialization: ; On ECU startup if ram location flag <> special value then ; Cold boot clear all RAM copy real time update tables from flash to RAM set ram location flag = special value else ; Warm boot clear RAM outside our tables end if

[0037] Most ECUs 15 have battery backed up RAM 105 so that RAM contents such as the idle learn parameter are retained whilst the vehicle battery 5 is connected via connection 120. For real time tuning using processing internal to the ECU 15, the RAM table memory 125 contents need to be retained over multiple driving cycles in order to give satisfactory self-tuning results. Upon initialization, the ECU erases all RAM, and the real time tuning code copies tables from flash memory to RAM as described above. For active tuning, it is desirable to retain the contents of the RAM tables between driving cycles. In order to do this, the ECU initialization code is modified so that RAM tables are only copied from flash memory from a cold boot (vehicle battery has been disconnected) but not from a warm boot (ignition has been switched on).

[0038] FIG. 3 illustrates a flow chart of the initialization process in the start-up of the system. In box 200, the ECU start-up occurs followed by decision box 205 in which it is determined whether this was a cold boot or not. That is, whether the battery 5 of the vehicle has been disconnected or the ECU memory contents has been otherwise erased. If the start-up is from a cold boot, the processor initiates the RAM in box 210 and then copies the data tables from the flash memory to the RAM in box 215. If the start-up is not a cold boot, the processor initializes the RAM memory without copying the tables from flash memory in box 220, and then the program exits in box 225.

[0039] The state of the real time tables may be saved between tuning sessions by external tuning software running on a laptop. If a tuning session is suspended and then resumed, then the external software can determine that live tuning for a particular table is active, and sends commands to the ECU to set the RAM table values without needing to re-write the whole ECU flash memory. This process takes a fraction of a second compared to minutes to re-write the ECU flash memory.

[0040] When using an ECU with real time tuning (as with the above method or any other method that allows the ECU memory to be altered whilst the ECU is running) and while using feedback from a wideband lambda sensor, the ECU fuel tables using the present invention can be altered by software in a closed feedback loop to bring the ECU fuel tables closer to optimum. The software uses parameters to determine when fuel table changes should be made, such as engine conditions (like temperature, throttle setting), lambda parameters (minimum & maximum readings to consider) and dynamic parameters (delay after engine events etc). A copy of the base fuel table is kept so that a maximum change in the fuel table can be specified. Additional parameters determine how often changes should be made, and what closed loop strategy is used (e.g., PID algorithms), and what table smoothing algorithm should be used.

[0041] As set forth above, the fuel table is a matrix of discrete fuel values. Engine parameters are scalar values that index each axis (two to three axes or more). Because the fuel table has discrete values, an algorithm must be used to determine which cell or cells should be altered. Options include: a) closest cell, where the closest matching fuel table cell to the indexes is altered; b) weighted cell, where the four closest cells (or eight cells for three axis tables) are altered by proportion of the cell midpoint to the index scalar value; and c) precise cell, where the software only changes a cell when the index values are close to the center of the cell.

[0042] In addition, all modern ECUs 15 have a knock sensor 65 and associated processing circuitry, that use engine noise processing to determine if engine knock is occurring, and optionally how close the engine 85 might be to engine knock. An additional ignition trim table may be created that allows the main ignition tables to be altered in order to minimize engine knock where knock occurs, and optionally increase engine timing when conditions allow it. Also, there are multiple algorithms available for tuning ignition tables: a) retard on knock, when the ignition is reduced in a cell by a fixed amount when there is knock present; b) adaptive retard when the ignition is also advanced back in a cell if no knock occurs; and c) noise based timing when the knock sensor noise (rather than the presence of knock) is used to trim ignition timing.

[0043] The active knock control routines are typically executed in a timer interrupt at a fixed frequency, such as 100 Hz. The active_knock_adjustment table is a 2 or more axes lookup table with rpm and engine load indexes containing ignition timing adjustment values used to trim the ignition timing. The reference_engine_noise table below contains typical knock sensor noise value when engine knock is not occurring.

[0044] To employ a knock count method into the active ignition tuning, a pseudo code such as the following can be used:

TABLE-US-00002 set step_size = user setting for knock retard step calculate rpm & load indexes for table lookup read knock_adjustment_value from active_knock_adjustment table set knock_count = sum of all cylinder knock count if knock_count <> previous_knock_count then previous_knock_count = knock_count set new_knock_adjustment_value = knock_adjustment_value + step_size if new_knock_adjustment_value < settings_maximum_knock_adjustment then update active_knock_adjustment table value with new_knock_adjustment_value end if else if using adaptive knock control method then set new_knock_adjustment_value = knock_adjustment_value step_size if new_knock_adjustment_value > settings_minimum_knock_adjustment then update active_knock_adjustment table value with new_knock_adjustment_value end if end if

[0045] In the pseudo code above (see FIG. 4), the step size is set to the value for the ignition retard when knock is detected in the engine in step 300. In the following step 310, a subroutine (discussed below) calculates the table indices using the table index values, typically rpm and load, The current ignition adjustment value is read in step 320 from the knock adjustment table using the table indexes. The program in step 330 sums the knock count for each cylinder, and compares the result with the previous knock count sum in step 340. If the knock count has increased in step 350, then the knock count adjustment value is increased by the step size up to the maximum knock adjustment value in step 360. If an adaptive knock method is used and there is no knock, then the knock adjustment value is decreased by the step size in step 370, where the adjustment value is limited by a minimum knock adjustment value, whereupon the program exits in step 380.

[0046] Alternatively, the active tuning can be used with a knock level method as well:

TABLE-US-00003 set knock_noise = knock sensor noise calculate rpm & load indexes for table lookup read knock_adjustment_value from active_knock_adjustment table read reference_engine_noise_value from reference_engine_noise table set adjustment = (knock_noise reference_engine_noise_value) * adjustment factor set new_knock_adjustment_value = knock_adjustment_value + adjustment if new_knock_adjustment_value < maximum_knock_adjustment and new_knock_adjustment_value > minimum_knock_adjustment then update active_knock_adjustment table value with new_knock_adjustment_value end if

[0047] In the above code sequence, referring to the flow chart of FIG. 5, the knock sensor noise is read in step 400, and the table indexes are calculated in step 410 using the table index values 415, typically rpm and load. The current ignition adjustment value is read in step 420 from the knock adjustment table 425 using the table indexes, and the reference engine noise is also obtained from the engine noise table 435 using the table indexes. An adjustment is calculated in step 440 from the difference between the reference noise and the knock sensor noise multiplied by the adjustment factor, and the adjustment is applied to the current knock adjustment value in step 450. A comparison of the new value is made in step 455, and if the new knock adjustment value falls within valid predefined adjustment value limits, then the adjustment value is saved to the table in step 460. Otherwise, the new value is discarded in step 470.

[0048] The table below illustrates an example of an ignition adjustment matrix after driving:

[0049] To modify the fuel tables using the ECU alone, a closed loop feedback routine is established. All modern ECUs use a wideband type oxygen sensor that reads the exhaust lambda ratio. Analyzing the measurements from this sensor can be used to trim the fuel tables using a closed loop feedback routine, as set forth below. The first step is establishing a target lambda, which is typically the stoichiometric ratio (lambda=1) for light and moderate load areas, which enrichment high load and cold operating temperatures. A sample target lambda table is set forth below:

[0050] Once the target lambda is established, the measurements from the oxygen sensor are compared with the target lambda to determine if the engine is operating at or close to the target lambda. The look-up table value is adjusted and the value is replace in the table when the adjusted value is greater than the previous value. In this way, the successive table values are continuously increasing.

[0051] The parameters used to control fuel active tuning are:

[0052] a. operational area (closed loop and/or open loop)

[0053] b. operational conditions (temperature, load, engine speed, gear etc)

[0054] c. operation limits (maximum trim)

[0055] d. correction methods (using existing ECU fuel trims)

[0056] e. transitional delays (engine start delay, injector restart delay)

[0057] f. sensor processing (sensor correction, lambda reading delay)

[0058] FIG. 6 illustrates an interface template that can be used to adjust the active tuning. The first box 500 is checked by the user when the active tuning is enabled. Below the activation box is an options group 510 that allows for open loop, closed loop, use corrected for wideband, and use fuel trim. Each option has a box that can be checked by the user. Below the options group 510 is a settings group 520, that require values to be input such as tuning method, delay after engine start, delay after injector re-start, change per tuning cycle, delay in lambda reading, update frequency, maximum change, maximum cam angle delta, maximum load delta, and maximum rpm delta. These user settings allow the program to operate under specified conditions and parameters. Below the settings group 520 is a conditions group 530, that includes rpm, load, TPS, VTC, ECT, IAT, and min/max lambda values. It should be noted that the lambda as read by the lambda sensor is not accurate for a short period after the fuel injectors, re-start after being shut-off, so a configurable delay is included to prevent the active tuning process until the engine lambda reading has stabilized. This can be seen by the chart in FIG. 7 illustrating the delay after the fuel injectors restart before the lambda sensors recognize the event.

[0059] FIG. 8 illustrates a graph in which the lambda sensor is delayed from the actual lambda due to the lambda sensor being in the exhaust system, downstream from the engine, and the system delay of the lambda sensor controller. This delay can be estimated by observing the time difference between engine events that result in a change of lambda and the change in lambda. For example, one method is to measure the time between injector shut off and the lambda reading changing.

[0060] To compensate for the delay in lambda reading when the ECU is in active tuning, a circular buffer is kept containing engine conditions used to index changes to the fuel adjustment table (rpm, load, cam position etc). When active tuning changes are made, the historical engine conditions are then used to determine where the fuel tuning adjustment should be made. The active fuel tuning routines are typically executed in a timer interrupt at a fixed frequency, such as 100 Hz.

[0061] When using a circular buffer routine, active tuning can be accomplished by the following pseudo code:

TABLE-US-00004 ; Implements a circular delay buffer for engine conditions to compensate ; for oxygen sensor delay set old_rec_no = engine_rec_ptr + 1 ; The size of the circular array is the minimum of the maximum size and the ; delay setting (at 100Hz the units are 10ms) set end_engine_rec_no = maximum buffer size if end_engine_rec_no > setting_lambda_delay then end_engine_rec_no = setting_lambda_delay end if ; Cope with circular buffer rolling over if old_rec_no = end_engine_rec_no then old_rec_no = 0 end if get engine conditions record using index old_rec_no call active fuel tuning conditions checking routine (below) ; Update circular buffer of engine conditions save engine conditions record at index engine_rec_ptr set engine_rec_ptr = engine_rec_ptr + 1 if engine_rec_ptr = end_engine_rec_no then engine_rec_ptr = 0 end if

[0062] In the foregoing code, an array of engine conditions are created in a memory, such as rpm, load, cam position, and the like obtained at a fixed frequency (100 Hz). Referring to FIG. 9, the oldest record number is set to the current record number in step 600, and the end record number is set to the lesser of the array element count and the lambda delay in step 610. If the old record number equals the end record number, then the buffer has rolled over so the old record number is set to the first element of the array in step 620. Then the engine conditions are set from the record for the array at the old record position in step 630. The subroutine set forth below is then run for active tuning in step 650. The current engine conditions are saved to the array at the current record position and the current record position is incremented by one in step 660. If the current record position has reached the end record number, then set the current record position to the first element of the array.

[0063] Similarly, active fuel tuning conditions using a checking routine:

TABLE-US-00005 set tps = throttle position set vtc = intake camshaft position set tps_ok = tps > setting_min_tps and tps < setting_max_tps set vtc_ok = vtc > setting_min_vtc and vtc < setting_max_vtc set engine_fuel_cut when injector are shut off ; Start a delay counter if the injectors are off, or the tps ; or cam angle is out of range if engine_fuel_cut or not tps_ok or not vtc_ok then set injector_delay_counter = setting_fuel_restart_delay else if injector_delay_counter > 0 then set injector_delay_counter = injector_delay_counter 1 else ; Table update frequency if active_tuning_counter > 0 then set active_tuning_counter = active_tuning_counter 1 else ; Check engine conditions set load_ok = load > settings_min_load and load < settings_max_load set rpm_ok = rpm > settings_min_rpm and rpm < settings_max_rpm set ect_ok = ect > settings_ min_ect and ect < settings_max_ect set iat_ok = iat > settings_min_iat and iat < settings_max_iat set gear_ok = gear > settings_min_gear and gear < settings_max_gear set time_ok = engine running time > settings_min_engine_run_time if engine is in closed loop then set cl_ok = setting_active_closed_loop else set cl_ok = setting_active_in_open_loop end if set lambda_ok = lambda > settings_min_lambda and lambda < settings_max_lambda if load_ok and rpm_ok and ect_ok and iat_ok and gear_ok and time_ok and cl_ok and lambda_ok then call active fuel tuning update adjustment table routine (below) end if ; Set the cycle delay counter set active_tuning_counter = settings active fuel update frequency end if end if

[0064] In this code sequence, illustrated by the flow chart of FIG. 10, throttle position, the camshaft position, and the injector overrun shut off are all checked to make sure they are within acceptable limits in step 700. If not, a delay counter is set to the setting for the fuel restart delay in step 710 and the routine is exited in step 750. If the delay counter is greater than zero, then the delay counter is decremented in step 720 and the routine is exited. If the active tuning frequency counter is greater than zero, then the active tuning frequency counter is decremented in step 730 and the routine is exited. If load, rpm, ect, iat, gear, after start delay, closed loop & lambda are within the established range, then the routine below is performed in step 740 and the active tuning frequency counter is set to the setting for tuning frequency.

[0065] The active fuel tuning can also update the adjustment table using pseudo code:

TABLE-US-00006 ; Find target lambda read target_af from target_af_table ; Adjust target lambda for closed loop operation (short & long term trims) if settings_use_fuel_trims then set adjustment = (short_term_fuel_trim + long_term_fuel_trim) set target_af = target_af* adjustment end if ; Find adjustment table indexes call find closest table indexes (below) ; Check the delta cam, load & rpm and within range if cam_angle_delta < settings_max_delta_cam and rpm_delta < settings_max_rpm_delta and load_delta < settings_max_load_delta then ; Calculate adjustment read current lambda adjustment value from table if lambda < target_af then ; Lean set delta_lambda = target_af lambda set adjustment = delta_lambda * fuel change per cycle set new_value = value + adjustment if new_value > settings_max_delta then new_value = settings_max_delta end if else ; Rich set delta_lambda = lambda target_af set adjustment = delta_lambda * fuel change per cycle set new_value = value adjustment if new_value < settings_min_delta then new_value = settings_min_delta end if end if update lambda adjustment table value with new_value end if

[0066] In the preceding routine, the target lambda is read from the target lambda table in step 800 of FIG. 11. If the setting to use fuel trims is set, then the target lambda is adjusted by the short and long term fuel trims in step 810. Using the routine below, the table indexes are calculated using the table index values, typically rpm & load, in step 820. If the rpm, load or cam angle delta values are greater than the setting for the maximum rpm, load and cam angle deltas then exit the routine in step 830. The current lambda adjustment value is read from the table, and a new adjustment value is calculated in step 840 as the difference between the current lambda and target lambda, multiplied by the setting for fuel change be cycle, and added to the old adjustment value. If the new adjustment value is within the setting for maximum adjustment, then the lambda adjustment table is updated with the new value in step 850.

[0067] To find the closest table index, the following code is used:

TABLE-US-00007 ; Find closest cam angle index ; cam_index_lookup is the table lookup scalar index if no vtc then ; use first cam angle table set cam_index = 0 set cam_angle_delta = 0 else set cam = engine intake camshaft advance angle set cam_index = number of cam angle tables 1 set cam_delta = 0 while index > 0 set mid_point = (cam_index_lookup[cam_index] + cam_index_lookup [cam_index 1]) / 2 set cam_angle_delta = cam cam_index_lookup[cam_index 1] if cam > mid_point then set cam_angle_delta = cam_index_lookup[cam_index] cam exit while loop end if set cam_index = cam_index 1 end while if cam_delta < 0 then set cam_delta = cam_delta end if end if ; ; Find closest rpm and load indexes ; rpm_index_lookup and load_index_lookup are the table lookup scalar indexes ; ; Rpm set rpm_index = rpm_index_lookup size 1 set rpm_delta = 0 while rpm_index > 0 set mid_point = (rpm_index_lookup[rpm_index] + rpm_index_lookup [rpm_index 1]) / 2 set rpm_delta = rpm rpm_index_lookup [rpm_index 1] if rpm > mid_point ; Rpm is higher than the mid point set rpm_delta = rpm_index_lookup[rpm_index] rpm exit while end if set rpm_index = rpm_index 1 end while if rpm_delta < 0 then set rpm_delta = rpm_delta end if ; Load set load_index = load_index_lookup size 1 set load_delta = 0 while load_index > 0 set mid_point = (load_index_lookup[load_index] + load_index_lookup[load_index 1]) / 2 set load_delta = load load_index_lookup[load_index 1] if load > mid_point ; Load is higher than the mid point set load_delta = load_index_lookup[load_index] load exit while end if set load_index = load_index 1 end while if load_delta < 0 then set load_delta = load_delta end if

[0068] In the foregoing routine, the rpm, load and current camshaft position are read in step 900 of FIG. 12. Then the index position is initialized in step 910 to the size of the cam angle tables. The mid point of the index values is calculated in step 920 as the average of the index value for the index position added and the index value for the next index position. If the current camshaft position is greater than the mid point, then the delta is recorded and the loop is exited in step 930. Otherwise, the index position is decremented in step 940 and the routine is looped. Calculating the rpm & load index routines are similar to the cam angle routine above.

[0069] Using the code above, the sample active tuned fuel adjustment table before driving found above is modified as follows after driving:

[0070] In an alternate embodiment, the airflow meter table is optimized during active tuning to determine the fuel requirements using mass flow rather than volumetric efficiency. The fuel injected quantity is based on information received from an air flow meter, where a look-up table is actively tuned for the air flow calibration instead of the volumetric efficiency tables of the previous embodiment. In this embodiment, the look-up table involves a single dimension value so the calculations are simpler than the preceding example.

[0071] For the Air-Fuel Meter active tuning, the code to determine the target lambda is still used, but a new update routine is employed. Rather than using an adjustment table like the speed density active tuning, when using the Air Fuel Meter active tuning the original table values are kept in another table and used to limit the changes to the AFM look-up table as per the parameters of FIG. 6.

[0072] The AFM update routine, the flow chart of which is shown in FIG. 13, is:

TABLE-US-00008 set AFM_voltage = time delayed value of the AFM voltage from the circular buffer set AFM_table_index = the AFM table index value for AFM_voltage set original_value = AFM_original_values[AFM_table_index] set limit_value = original_value * max delta set current_value = AFM_table[AFM_table_index] set new_value = current_value if lambda < target_af then ; lean set delta_lambda = target_af lambda set adjustment = delta_lambda * fuel change per cycle if adjustment < 1 then adjustment = 1 if current_value + adjustment < limit_value then new_value = current_value + adjustment else ; rich set delta_lambda = lambda target_af set adjustment = delta_lambda * fuel change per cycle if adjustment < 1 then adjustment = 1 if current_value adjustment > limit_value then new_value = current_value adjustment ; make sure AFM table value increase if new_value < previous value in AFM table then new_value = previous value in AFM table if new_value > next value in AFM table then new_value = next value in AFM table

[0073] The AFM table values are checked when updating to ensure that successive values are larger than the preceding values. The same process could be used with the fuel injector table. That is, if only one variable is changed at a time on a stock vehicle, then the difference in engine operation can be attributed to that change and the corresponding look-up table is adjusted accordingly.