阅读说明：本技术 在考虑测量噪声下的用于测量飞行器发动机轴的转速的方法 (Method for measuring the rotational speed of an aircraft engine shaft taking into account measurement noise ) 是由 赛德瑞克·德杰拉希 雅克·保罗·米歇尔·高弗里特 让-弗朗索瓦·让-保罗·克莱芒·席林 于 2018-06-13 设计创作，主要内容包括：本发明涉及一种用于测量飞行器发动机轴的转速的方法,该方法包括以下步骤：获取跨音轮传感器的端子的交流转速检测信号；将所述交流信号转换为方波信号；将先前记录的多个方波信号周期样本与周期上限和周期下限进行比较,以确定其值在下限与上限之间的有效样本；如果有效样本的数目大于第一阈值,则基于所述有效样本确定方波信号的期望周期；以及至少在有效样本的数目小于第一阈值的条件下,计算样本的多个和,并计算包括有效周期样本与至少两个样本的多个和中的一些和的集合的平均值。(The invention relates to a method for measuring the rotational speed of an aircraft engine shaft, comprising the following steps: acquiring an alternating current rotating speed detection signal of a terminal of a transonic wheel sensor; converting the alternating current signal into a square wave signal; comparing a plurality of previously recorded square wave signal period samples with an upper period limit and a lower period limit to determine valid samples having values between the lower limit and the upper limit; determining an expected period of the square wave signal based on the valid samples if the number of valid samples is greater than a first threshold; and calculating a plurality of sums of the samples and calculating an average of a set including the valid periodic samples and some of the plurality of sums of the at least two samples, at least on a condition that the number of valid samples is less than a first threshold.)

1. For generating a square wave signal (S)_C) Method for measuring the speed of rotation of an aircraft engine shaft, said square-wave signal being an alternating speed detection signal (S) obtained from terminals of a speed sensor_V) And a speed sensor including a tone wheel rotationally driven by a shaft,

the square wave signal (S)_C) Corresponding to a restart threshold value (U) being lower_R) And above a reset threshold (U)_A) Ac signal (S)_V) Are respectively switched to a low level (U)₁) And high level (U)₂)，

The method comprises the following steps:

storing square wave signals (S) in a memory of circular buffer type (B)_C) A plurality of periodic samples (T)₁，…，T_n)，

Each of the samples is associated with a lower period limit (P)_tmin) And the upper limit of the period (P)_tmax) Comparing to determine a valid sample having a value between the lower limit and the upper limit; and

if the determined number of valid samples is greater than a first threshold (A)₁) Determining the square wave signal based on the valid samples (S)_C) Desired period (P) of_t)，

Said method being characterized in that, at least when the number of valid samples is less than a first threshold value (A)₁) In the case of (A), the square wave signal is determined by the following steps_C) Desired period (P) of_t)：

Calculating a plurality of sums (S) of at least two samples₁₂，…，S_1n，S₂₃，…，S_(n-1)n)，

Determining a valid sum of consecutive samples from the plurality of sums, the valid sum being at the lower limit of the period (P)_tmin) And the upper limit of the period (P)_tmax) And are each independently of the other

Calculating an average value of a set comprising a valid sum of a plurality of valid period samples and a plurality of consecutive samples, the desired period (P) of the square wave signal_t) The value is equal to the average value.

2. The method of speed measurement according to claim 1, the method further comprising the additional step of: determining at said upper and lower cycle limits (P) from said plurality of calculated sample sums_tmin，P_tmax) The sample valid sum in between; and

calculating a desired period (P) of the square wave signal by calculating an average of a set of the active period samples and the determined active sum of samples_t)。

3. The speed measurement method according to claim 1 or 2,

if the sum of the number of valid samples and the number of valid sums of samples is less than a second threshold value (A)₂) Said desired period (P) of said square wave signal_t) A value equal to the desired period (P) of the square-wave signal obtained during the previous iteration of the method_t-1)。

4. The speed measurement method according to any of claims 1-3, wherein the lower and upper period limits (Pmax) are calculated based on a fixed period value and a maximum period variation value_tmin，P_tmax)。

5. Measuring method according to claim 4, wherein the fixed period is taken to be equal to the desired period (P) of the square wave signal obtained at an earlier time_t-1)。

6. Measuring method according to any of claims 4 or 5, wherein the law of maximum rotational speed gradient (G) is used_N1max) To obtain a maximum period variation value, wherein the maximum gradient at a given time is a function of the rotational speed.

7. Measuring method according to any of claims 1 to 6, wherein, during an initialization phase of the measurement, when the number of valid periodic samples does not exceed a third threshold (A)₃) According to the period (P)_th) To estimate the expected period of the square wave signal.

8. Measuring method according to any of claims 1 to 7, wherein during an initialization phase of the measurement, when during the initialization phase the square wave signal (S)_C) Upon detecting that the mutual dispersion of the periodic samples exceeds a predetermined threshold (D), according to the period (P)_th) To estimate the expected period of the square wave signal.

9. Measuring method according to claim 7 or 8, wherein during an initialization phase the desired period of the square wave signal is determined by means of a numerical model according to a thermodynamic parameter of the engine, preferably selected from a high pressure rotor speed value, or a speed value of another rotating member of the engine, an atmospheric pressure value or a compressor stator level value.

10. An aircraft engine control unit configured to perform the steps of: by square wave signal (S)_C) As an input to the process, the process may,

storing the square wave signal (S) in a memory of circular buffer type (B)_C) A plurality of periodic samples (T)₁，…，T_n)；

Each of the samples is associated with a lower period limit (P)_tmin) And the upper limit of the period (P)_tmax) Comparing to determine a valid sample of values between the lower limit and the upper limit；

If the determined number of valid samples is greater than a first threshold (A)₁) Determining the square wave signal based on the valid samples (S)_C) Desired period (P) of_t) (ii) a And

if the number of valid samples is less than the first threshold value (A)₁) Determining said square wave signal (S) by_C) Desired period (P) of_t)：

Calculating a plurality of sums (S) of at least two samples₁₂，…，S_1n，S₂₃，…，S_(n-1)n)；

Calculating an average value of a set comprising some of the plurality of active period samples and a plurality of sums of the at least two samples, the desired period (P) of the square wave signal_t) The value is equal to the average value.

11. A computer program product for use in an application system of a turbine engine control unit, the code instructions of the computer program product being capable of implementing the method according to any one of claims 1 to 9.

Technical Field

The invention relates to accurate measurement of aircraft engine speed.

More specifically, the presence of spurious noise added to the desired signal in an unpredictable manner is considered in relation to the measurement of the rotational speed of the engine shaft by the tone wheel sensor.

Background

It is necessary to accurately detect the rotational speed of the aircraft engine shaft, since the detection of the rotational speed of the aircraft engine shaft can control the switching off or regulation of the engine power supply in the event of, for example, the rotational speed or the acceleration exceeding a predetermined safety value.

It is known to use a variable reluctance sensor associated with the tone wheel to measure the rotational speed of the shaft. The sensor may be fixedly referred to as a tone wheel sensor or a variable reluctance sensor. The tone wheel is arranged to form a single component with the shaft. The sensor uses a winding to generate a magnetic field that closes on the air gap between two teeth of the tone wheel or on one tooth of the tone wheel. The measured magnetic flux thus varies according to the passage of the teeth of the tone wheel in a manner proportional to the rotational speed of the tone wheel. The frequency of the alternating voltage thus generated is equal to the passing frequency of the teeth of the tone wheel, which itself describes the speed of rotation of the shaft, and the amplitude of this alternating voltage is a function of the air gap and of the signal frequency. The applicant, in international patent application No. WO 2014/207369, describes a toothed nut that can be screwed onto an aircraft engine shaft, the teeth of which can cooperate with a rotation speed sensor operating according to this principle, to mechanically fix the engine shaft, for example with ball bearings. Furthermore, at least a portion of the teeth of the nut also comprise cavities, providing a maximum number of gap intervals in this portion while maintaining the tightening function of the nut, so that the rotation speed of the engine shaft can be measured accurately.

The speed measurement system with a tone wheel can be used in particular for measuring the so-called "N1" speed of rotation of the low-pressure rotor of a twin-rotor jet engine. On a "direct drive" engine (fan driven directly by the low spool), this speed N1 is also the speed of the fan. The tone wheel sensor may also be used to measure the rotational speed N2 of the high pressure spool of a twin spool injection engine, or the rotational speed of another rotating member of the engine.

The sinusoidal alternating current signal taken across the terminals of the variable reluctance sensor may be converted into a square wave signal by an electronic circuit located upstream of the engine control unit, which may perform a frequency measurement on the square wave signal. The conversion of the sinusoidal signal into a square-wave signal is performed in particular by a schmitt trigger having a predetermined restart threshold and a reset signal. The square-wave signal at the output of the schmitt trigger switches to a high level when the input voltage exceeds a reset threshold and switches to a low level when the input voltage is below a restart threshold.

However, the spurious noise can severely interfere with the square wave signal provided to the speed sensor with the tone wheel. The components of the sensor are subject to electromagnetic interference due to mechanical vibrations of certain engine components. Intermittent noise events and an alternating-current type noise signal superimposed on the signal indicative of the rotation of the tone wheel (desired signal) can be observed. In this regard, fig. 1 shows, on the first coordinate axis, the voltage S across the tone wheel sensor terminals in the case where the signal is changed by an alternating-current type spurious noise (the frequency of which is greater than that of the desired signal)_VTime-varying and on the second axis the voltage signal S at the output of the flip-flop is shown_VSquare wave signal S obtained after squaring_C. The peak-to-peak voltage due to noise may be locally greater than the reset threshold U_RAnd reset threshold U_AAnd thus ac noise, can cause spurious switching of the square wave signal. Third stepThe axes show the square-wave signal S for successive periods_CSignal C for counting_PChange over time. The time delay shown by the curve of this third axis restarts with each down-switch of the square wave signal from high to low level. For example, in the time period P shown in the figure₁In the above, there is no spurious switching due to noise. On the other hand, in the next period P₂Due to parasitic ac noise superimposed on the desired signal, the detected signal S_VCrossing reset threshold U_AAnd then falls again to the restart threshold U_RThe following. The time between two passes triggering the threshold is indeed less than the period P₁This results in a square wave signal S_CAnd results in the detection of the square-wave signal S in the position of the single "complete" period P2 obtained in the case of non-spurious switching_CTwo successive "divided" periods P of_2aAnd P_2b. The rotational speed of the shaft is then disturbed.

Known solutions include filtering out spurious noise upstream of the electronic device generating the square wave signal. This is a low level solution implemented upstream of the engine control unit. For example, low pass filters are typically utilized to filter out noise having frequencies greater than the frequency of the desired signal. However, low pass filtering tends to cause a phase shift in the obtained measurement. If the position of the engine shaft is used as a reference to monitor other components of the engine, the measurement of the phase shift may interfere with other functions that may be performed by the signal acquired across the terminals of the variable reluctance sensor, such as engine balancing functions. In addition, low pass filtering may reduce the amplitude of the signal. With a constant restart threshold and reset threshold and with a decreasing amplitude of the rotation speed measurement signal, the accuracy of the obtained frequency measurement decreases. This problem is even more relevant at low engine speeds (e.g. at engine start-up), in which case the ratio of the peak-to-peak amplitude of the ac-type spurious noise to the peak-to-peak amplitude of the desired signal may be greater than at high engine speeds.

Furthermore, there are software solutions for post-processing the sinusoidal voltage acquired across the terminals of the variable reluctance sensor, which are implemented on the application system (for example, a FADEC type system) level of the engine digital interface. However, none of the existing solutions is well suited to deal with the problem of noisy signals due to high frequency ac noise.

In particular, the prior art solution, which determines the passing period of the teeth of the tone wheel at the present moment on the basis of a simple average of the period measurements taken at an earlier time, is only effective if a sufficient number of "complete" periods free of noise are obtained.

Disclosure of Invention

The present invention solves the above problem by post-processing the values of the detected periods on the square wave signal, which can be implemented at the level of the software application of the engine digital interface, to reconstruct the "expected period" (i.e. the period of the expected signal obtained in the absence of spurious noise) based on a set of periodic samples of the square wave signal stored in a memory of the circular buffer type. This post-treatment is based on the following observations: in the presence of spurious switching (e.g., signal S previously shown in FIG. 1)_CFalse switching of the up-emphasis) can be performed by taking the signal S for several consecutive cycles_CTo obtain a value very close to the observed period for the desired signal. The method of the invention enables the correct sum to be found by taking a plurality of sums of successive periodic samples and comparing these sums with a likelihood threshold calculated from previously taken periodic and rotational speed measurements.

According to a first aspect, the object of the invention is a method for measuring the rotation speed of an aircraft engine shaft, comprising the steps of:

acquiring an alternating current rotational speed detection signal acquired across terminals of a speed sensor, the speed sensor including a tone wheel rotationally driven by the engine shaft;

converting the alternating current signal into a square wave signal, and respectively switching the square wave signal to a low level and a high level corresponding to the passing alternating current signal voltage being lower than a restart threshold and the passing alternating current signal voltage being higher than a reset threshold;

storing a plurality of samples of the square wave signal period in a memory of the circular buffer type and comparing each of said samples with a period lower limit and a period upper limit to determine valid samples having values between these two limits;

determining an expected period of the square wave signal based on the valid samples if the determined number of valid samples is greater than a first threshold; and

determining an expected period of the signal by, at least on condition that the number of valid samples is smaller than a first threshold value:

a plurality of sums of at least two samples are calculated,

an average value is calculated comprising a plurality of valid period samples and a set of a plurality of sums taken from a plurality of sums of at least two samples, the desired period of the square wave signal being taken to be equal to the average value.

The method of the present invention need not have periodic samples corresponding to a complete period of the desired signal. Even with only "split" periodic samples due to spurious switching obtained on the square wave signal, the method of the invention is able to obtain a period close to that obtained for the desired signal.

The process of the invention is developed starting from the following observations: signal S as shown, for example, in FIG. 1 above_CIn the presence of false switching of the up-highlighting, the signal S can be obtained by taking a signal S of several consecutive periods_CTo obtain a period value that is very close to the period value observed for the desired signal.

Advantageously, the method may exhibit the following additional features:

the method comprises the additional steps of: determining a valid sum of samples between the upper limit of the period and the lower limit of the period from the calculated sums of the plurality of samples, calculating an expected period of the square wave signal by calculating an average value of a set consisting of the valid period samples and the valid sum of the samples;

samples that add together to determine a valid sum are consecutive samples;

setting the desired period of the square-wave signal equal to the desired period of the square-wave signal obtained during the previous iteration of the method if the sum of the number of valid samples and the number of valid sums of samples is less than a second threshold;

calculating a lower and an upper period limit from the fixed period value and the maximum period variation value;

then, the fixed period is set equal to the desired period of the square wave signal obtained at an earlier instant;

then, the maximum periodic variation value is obtained using the law of maximum rotational speed gradient, where the maximum gradient at a given moment is a function of the rotational speed;

during the following initialization phase of the measurement: estimating the expected period of the square wave signal by a theoretical numerical model of the rotating speed at a stage where the number of effective period samples does not exceed a third threshold value or at a stage where the mutual dispersion of the period samples detected on the square wave signal exceeds a predetermined threshold percentage;

then, during an initialization phase, the desired period of the square-wave signal is determined by means of a numerical model of the rotation speed according to a thermodynamic parameter of the engine, preferably a high-pressure rotor rotation speed value, an atmospheric pressure value or a compressor stator shim value.

According to a second aspect, the invention relates to a turbine engine control unit configured to implement the steps of the above method.

According to a third aspect, the invention finally relates to a computer program product intended for use in an application system of a control unit of a turbomachine engine, the code instructions of which enable the implementation of the above-mentioned method.

Drawings

Fig. 1 has been described above.

Other features, objects and advantages of the invention will become apparent from the following description taken in conjunction with the accompanying drawings which are meant to be illustrative only and not limiting.

Fig. 2 shows a plurality of false detections of square wave signal periods on an example of a noisy signal at the output of the tone wheel sensor.

Fig. 3 shows a theoretical voltage signal resulting from the superposition of a desired signal and sinusoidal ac-type noise to illustrate the principle of periodic sample addition.

Fig. 4 schematically shows an embodiment of an algorithm for processing period samples taken in a loop memory to recalculate the desired period.

Fig. 5 shows an embodiment of an algorithm for determining likelihood intervals in the algorithm of fig. 4.

Fig. 6 shows the results obtained on the basis of the signals relating to the speed of rotation of the shaft during a period of time comprising a limited time range Z during the initialization phase of the measurement, using the algorithm shown in fig. 4.

Fig. 7 shows the signal of fig. 6 on the short timescale of range Z.

Detailed Description

Throughout the description below, the terminals of the transonic wheel sensor acquire an alternating signal S_VFor detecting the rotational speed of the shaft of the low-pressure spool (BP) of a twin spool jet engine. Downstream of the sensor and upstream of an Electronic Engine Controller (EEC) (e.g., FADEC3), the signal S_VConverted into square wave signal S_C. For this purpose, the signal S is here_VFiltered by an RC filter, then peak-removed, then filtered again, finally by a restart threshold U with 0V_RAnd a reset threshold U of 0.232V_AThe schmitt trigger of (a) is squared.

Measuring a square-wave signal S by taking into account the time elapsed between two falling edges of the square-wave signal_CThe passing period of the teeth of the upper tone wheel. Here, the tone wheel sensor measures the engine speed N1. However, the same process conditions may be advantageously used for measuring any other rotational speed of the rotating member of the engine, for example measuring rotational speed N2.

However, as can be seen above with respect to fig. 1, the square wave signal S_CA kind of switching may be present which does not correspond to the rising or falling edge of the desired part of the rotation speed detection signal, but in contrast is caused by a voltage variation of the parasitic alternating current noise (undesired part of the signal). This phenomenon is illustrated in fig. 2. On the first axis, the theoretical voltage S of the terminals of the transonic wheel sensor_VRepresented as time tA function. And also shows the signal S_UBy a desired portion of the slave signal S in a theoretical manner_VMinus the contribution of parasitic ac noise. On a second time axis, the sum signal S is superimposed_VCorrelated square wave signal and theoretical desired signal S_UCorrelated square wave signal at low level U₁And a high level U₂To switch between. When the signal S_VIs above a threshold value U_AThen, these square wave signals are switched to high level U₂And when the signal S is_VIs reduced to a threshold value U_RThese square wave signals are switched to the low level U in the following₁。

Observed on the second axis, after squaring, is a noisy signal S_VHaving rising and falling edges corresponding to the portions 20', while after squaring, the signal S is expected_UWith a single rising edge and a single falling edge corresponding to section 20. Furthermore, the rising and falling edges of section 20' do not correspond to the rising and falling edges of section 20. Thus, with respect to the desired signal S_UObtained period for noisy signal S_VThe detected period is erroneous (e.g. the period between two consecutive falling edges of the square wave signal). However, the desired signal S_URepresents the desired speed N1.

The following description applies to square-wave signals S_CA plurality of period values T obtained above₁，…，T_nThe processing method of (1). The method 10 is implemented by an algorithm implemented in an application of an Electronic Engine Controller (EEC). The method is used to determine the value T₁，…，T_nTo reconstruct the fixed period P_t. Again, here, between two consecutive falling edges of the square wave signal (from the high voltage level U)₂To a low voltage level U₁) Obtaining a square wave signal S_CThe period of (c). It is not necessary to assume in advance that these periods are "complete", i.e. related to the theoretically expected speed detection signal excluding spurious noise, or "split", i.e. in the noise-related signal S_CIs calculated between spurious handovers. Method 10 for reconstructingt periods P_tThe period providing the theoretical desired signal S_UIs the best approximation of the period of. The method 10 is implemented digitally at the application software level of the EEC digital interface of the aircraft engine.

Fig. 3 is used to explain the purpose and principles of the method 10 using a simplified theoretical example. On a first axis, a time-dependent voltage signal S is shown_VExample of the voltage signal S_VBy applying a low-frequency, sinusoidal theoretical desired signal S, also represented on the first axis_UAnd a high frequency noise signal. The first axis also shows the signal S_VAnd theoretical signal S_URestart threshold U of a squared Schmitt trigger_RAnd reset threshold U_AThe S of_VAnd S_UThese two signals correspond to a square wave signal S represented horizontally on a second axis_CAnd a theoretical square-wave signal S represented horizontally on a third axis_CT. It can be seen that for the signal S_CTObtained period P_tIn a first period T₁To and from signal S_CThe periods effectively obtained partially overlap and then follow two periods T₂And T₃Completely overlapping and partially overlapping during the fourth non-reference period of fig. 3.

However, by applying successive periods T₁，T₂And T₃Adding up to obtain a very close value P_tThe value of the period of (P), the_tThe values constitute the desired period for the frequency measurement. One explanation for this result is that, given period P_tIn the case of (2), the periods T are respectively marked₁Start and period T of₃S of (2)_CCorresponds approximately to the signal S_CTA continuous falling edge is observed. If an analogy is made between this theoretical example and the actual measurement of the detection signal collected across the terminals of the transonic wheel sensor, the signal effectively observed across the terminals of the sensor will be the signal S_VSo that the resulting square wave signal is S_C. It is therefore an object of the method 10 to obtain a square wave signal S_CTThe resulting period P_tThe best approximation of, the period P_tCannot be observed directly.

The method 10 for processing a square-wave signal S is described next_CThe period value of the square wave signal S_CIs a rotational speed detection signal S obtained from a terminal of a transonic wheel sensor_VIt is contemplated that method 10 may be implemented by a digitally interfaced application software program or system of an electronic engine controller. The steps of method 10 will be described hereinafter with reference to fig. 4.

In step 100, a periodic sample value T of a square wave signal is obtained in the following manner₁，…，T_n. The first counter detects a falling edge of the high frequency square wave sampling signal related to the desired frequency of the square wave signal. For example, a measuring plate of the passing period of the tone wheel teeth may have a value of about 0.25.10^-6s sample periods of the temporal order. The second counter detects the square wave signal S_CThe falling edge of (c). Thus, the square wave signal S can be converted into a square wave signal_CIs stored in a circular buffer type of circular memory B, in which a certain number of cycle measurements can be stored. Here, it is preferable to implement buffer B for period measurement in an electronic circuit, rather than digitally in an application system of a digital interface of an electronic engine controller, since the calculation period of the engine controller is typically on the order of 0.015s, which is not capable of accurately measuring the passing period of the tone wheel teeth (for 60 teeth, the speed may reach 7000 revolutions per minute). The circular buffer B may then store the value T₁，…，T_nAnd the date of measurement of these samples is provided to the application system.

The number of samples (denoted here as n) is considered to be greater than the minimum number of samples required to ensure good measurement accuracy. The minimum number may for example correspond to the number of teeth of the tone wheel. By averaging the number of consecutive samples greater than the number of teeth of the tone wheel in method 10, repeated hereinafter in the subsequent steps, any differences between the measurements relating to the non-uniformity of the geometry of the tone wheel teeth are eliminated. In determining the minimum number of samples per buffer B, it is also referred to the expected frequency of spurious noise whose influence is desired to be reduced: the higher the noise frequency relative to the desired signal, the more "split" periods may be required to reconstruct the desired pass period of the tone tooth, according to the principles described with respect to fig. 3.

Then, in step 110, the samples T are aligned in order from the oldest to the newest sample₁，…，T_nAnd (6) sorting. This step 110 makes it possible to take the sum of successive cycles over time in subsequent steps. For convenience, it is assumed herein that the sorted samples retain their slave T after sorting₁To T_nThe order of (a). Thus, at the output of step 110, the application has a vector of n consecutive cycles of samples.

In the present embodiment, after step 110, there are two different scenarios depending on the value of the binary initialization signal indicating whether the period measurement value is in the initialization state.

Specifically, as described below, in the following step of reconstructing the fixed period based on step 120, the fixed period value obtained at the previous calculation increment is taken as an input. Therefore, these steps require at least one obtained fixed period value that is considered satisfactory.

In the present embodiment, the periodic samples T sorted in the buffer B are sorted₁，...，T_nAnd (6) carrying out testing. Alternatively, the initialization test may also be performed on unsorted periodic samples. Test specimen T₁，…，T_nIs sufficiently correlated, i.e. if their values are weakly dispersed with respect to a predetermined percentage of the dispersion threshold D.

For example, if the initialization test requires that three periodic samples be weakly dispersed from each other and for a dispersion threshold of 5%, a check is made to verify that the logical equation is satisfied

And

alternatively, the initialization test may be performed after verifying that the likelihood interval [ P ] is included_min，P_max]T between₁，…，T_nWhether the number of periodic samples in (a) exceeds a predetermined threshold a₃Wherein P is calculated using a theoretical model dependent on the operating conditions of the engine, for example_minAnd P_max。

As long as the initialization test does not give a positive result for the iteration of method 10, the value of the binary initialization signal will be zero and the measurement is considered uninitialized. In an initialization phase, in which the measurement has not yet been initialized, not from the previous fixed period P_t-1To determine the fixed period P_tBut by means of a theoretical numerical model P in step 500D (mode D of the determination step 500 of the period Pt)_thTo calculate P_t. The theoretical numerical model P is based on thermodynamic parameters of the engine (such as the rotation speed of the high-pressure rotor, the atmospheric pressure or the shim of the compressor)_thMay correspond to a theoretical model of the value of the speed of rotation N1. The results obtained during the initialization phase of the periodic measurement will be described below with reference to fig. 6 and 7.

If the initialization test result is negative for an excessively long period of time (relative to the predetermined time), which means that the signal is very noisy, the value of the binary initialization signal can be changed to 1 and the fixed period P is used_tThe fixed period P to initialize the algorithm_tIn a given period P_thIn the case of (2) is obtained from a theoretical numerical model. Due to the uncertainty of the measurement of the engine speed used in the theoretical model, the likelihood interval P is established in a subsequent iteration after initialization_{t min}，P_{t max}]The range of errors may be taken into account when determining the theoretical period.

If the initialization test results in a positive result, the binary initialization signal will increase to a value of 1. Then, until the end of the measurement of the passage period of the tone-wheel teeth, the initialization signal remains equal to 1 and, during the subsequent iterations of the method, the fixed period previously obtained is usedA period value. During an iteration of the method, in which the value of the binary initialization signal increases to 1, an initial fixed period P may be calculated from samples determined to be mutually coherent_t0. For example, the value P_t0May be taken to be equal to the arithmetic average of periodic samples that are considered to be mutually coherent. Then, the subsequent step 120 may be performed in subsequent iterations, considering the previous fixed period value as the value P_t0。

With the periodic measurements initialized, the method 10 continues with a step 120 of calculating lower and upper limits of the desired period after the step 110 of ordering the samples. Alternatively, step 120 may be performed before step 110, or in parallel with step 110, when measuring at an initialization period. Step 120 consists in establishing a likelihood interval which may then be at the ordered sample T₁，…，T_nWhich may correspond to a "complete" period, in other words a period sample valid for measuring the passing frequency of the tone tooth.

The calculation of the tth period P for the passage of a tone tooth is described in connection with FIG. 5_tBy a lower bound P, and the details of the algorithm 120 of the likelihood interval of_{t min}And an upper limit P_{t max}And (4) defining.

In step 121, the values P are obtained separately_t-1And V_t-1Of the P_t-1And V_t-1Are the (t-1) th value (previously determined using algorithm 100) and the (t-1) th corresponding rotational speed value, respectively, of the tone wheel tooth pass cycle. The period is expressed in milliseconds, for example, and the rotational speed is expressed in revolutions per minute (rpm), for example.

Next, in step 122, a predetermined law for the maximum possible gradient of the tone wheel rotational speed (maximum gradient of the rotational speed N1) is used to calculate the period P for_tMaximum rotational speed gradient. Law G_N1maxMay be from engine testing or previous calculations. Law of motion G_N1maxHere expressed as a function of the rotational speed of the shaft, but may also depend on other parameters related to the operating conditions of the engine. The speed gradient may be expressed in rpm per second.

In step 123, the value G is obtained_N1max(V_t-1) Is used to obtain the period P_tThe maximum rotational speed of the rotor. Here, the maximum gradient value G is set_N1max(V_t-1) Multiplied by the period P_t-1To obtain an estimate of the maximum variation. The maximum variation value here is an absolute value and may cause acceleration or deceleration of the tone wheel rotation. Thus, the velocity V_{t max}By adding the maximum variation to V_t-1Is obtained by the above calculation, and the velocity V_{t min}By passing from V_t-1The maximum variation is subtracted to calculate the maximum variation.

Alternatively, the maximum amount of change in the deceleration (relative minimum) direction of the rotation speed may not be equal to the opposite of the maximum amount of change in the acceleration direction.

Finally, in step 124, the maximum speed V is determined_{t max}And minimum velocity V_{t min}Of these values, the minimum period P is respectively deduced_{t min}And a maximum period P_{t max}The value of (c). First, according to V_{t max}And V_{t min}Respectively obtaining the maximum value and the minimum value f of the passing frequency of the tone wheel teeth_{t max}And f_{t min}. For example, if the speed is expressed in rpm, the speed expressed in rpm is first divided by 60 (to obtain a rotational speed value per second), and then the value obtained after the division is multiplied by the number of teeth of the tone wheel to obtain the maximum frequency and the minimum frequency at which the teeth pass. Finally, to obtain the corresponding period P_{t min}And P_{t max}In the case of adding saturation values in a known manner to avoid division by 0, the frequency f is taken into account separately_{t max}And f_{t min}The reciprocal of (c).

Returning to the schematic diagram of FIG. 4, in step 200, according to the period P_{t min}And P_{t max}Limited likelihood interval at T₁，…，T_nIn the period P_{t min}And P_{t max}Between significant samples T₁，…，T_k1And the number k of these valid samples₁(card V₁). Here, V₁Representing the set of periodic samples in buffer B designated as valid at this stage.

Then consider for valid samplesNumber of first threshold values A₁. Number k of valid samples (not necessarily consecutive)₁And a threshold value A₁A comparison is made.

If k is₁Is greater than or equal to A₁It is estimated that enough valid samples are obtained to obtain the tth period P of the tone tooth passage by averaging (here, arithmetic mean, or geometric mean) the valid samples_tThis is of great importance for the measurement of the rotational speed N1. In other words, among the samples contained in buffer B, there are enough samples that are estimated not to come from a spurious switch but to correspond to a "complete" cycle. Therefore, if k₁Is greater than or equal to A₁Then the application system of the electronic engine controller determines P in step 500A (mode a of step 500)_tIs equal to k contained in B₁The arithmetic mean of the samples of one active period.

Selecting a threshold A₁The threshold value A₁It is sufficient to obtain a fixed period of importance for the frequency measurement, which is not too high in order to make the calculation too complex and overload the memory of the application system. For example, consider the threshold A₁It is advantageous to be equal to the number of teeth of the tone wheel or a multiple of this number. Thus, if some teeth exhibit geometric irregularities relative to others, i.e. demonstrate that the periods in the desired signal are not equal (independent of the presence of spurious noise), the algorithm will reduce the effect of these irregularities by averaging over a number of periods covering the entire circumference of the tone wheel.

On the other hand, if k₁Is less than A₁Then the calculation of the sample sum of the next step 300 needs to be performed.

As previously seen with respect to fig. 3, one principle of adding several invalid period samples is to obtain a sample sum corresponding to a "complete" period of the desired signal by eliminating the effects of spurious switching caused by noise. In practice, the algorithm seeks the sum of consecutive samples, the value of which is included in the likelihood interval previously determined in step 120, i.e. at P_{t min}And P_{t max}In the meantime.

Thus, in the embodiment described here, the algorithm calculates the sum S for index i (i is from 1 to n-1, n being the number element of buffer B) and index j (j is strictly greater than i and from i +1 to n)_ijFrom T of the representation_iTo T_jOf buffer B, note the sum of the periodic samples in sample T₁，...，T_nHas been sorted in chronological order in 110. In other words, here all possible sums of consecutive samples in buffer B are performed.

Next, in step 400, at S₁₂，…，S_1n，S₂₃，…，S_(n-1)nAnd is based on a period P_{t min}And P_{t max}Defining a defined likelihood interval to determine the sample S₁，...，S_k2(the value of the sample is included in the period P_{t min}And P_{t max}In between) and the number k of valid sums of these samples₂. Here, V₂Representing a set of valid sums that are considered samples. Alternatively, values of the lower limit and the upper limit of likelihood values different from the value determined in step 200 may be considered, or alternatively, values belonging to the interval [ P ] may be applied_tmin，P_tmax]Different other criteria to select the valid sum, or may not select the valid sum.

In a manner similar to step 400, the number k may be based on the number of valid sums for a cycle₂(card V₂) And a second threshold value A₂The arrangement relationship between them distinguishes the two cases. Threshold value A₂May be taken to be equal to A₁. Preferably, the threshold value A₂Is taken to be greater than A₁Such that the fixed period P is calculated by validating and averaging the samples_tA greater number of effective sums are required. This is reasonable because it allows the use of a sample sum (rather than just a periodic sample) in a fixed period of calculation to reduce the accuracy of the measurement.

If the number of sample valid sums k₂Greater than or equal to threshold A₂Then in step 500B the algorithm determines a fixed period P_tIs equal to k₁One sample (k previously determined to be valid in step 200)₁One sample) and k₂A sample valid sum (refer to k of step 400)₂Sample valid sums) of the sets (here, arithmetic).

Unlike the calculation of step 500A, this last calculation of a fixed period is highly advantageous in that it does not require a predetermined number of valid samples in buffer B alone. In other words, the calculation of step 500B may be performed even if the set of acquired periodic samples or the large number of periodic samples in buffer B are "split" and caused by spurious switching. The algorithm reconstructs the desired period for the rotation speed calculation based only on the incoherent period (taking into account the rotation speed gradient law used), which not only enables an improvement in the accuracy of the filter output measurement, but also even in the case of square-wave signals S_CThe algorithm may also run without collecting "complete" cycles.

However, based on the invalid samples contained in buffer B, one successful condition for the fixed period calculation of step 500B is to reconstruct the sum of consecutive period samples determined to be valid, since these valid samples are included in likelihood interval P_{t min}And P_{t max}In the meantime. Preferably, if this condition is not met (i.e. if the periodic samples taken in buffer B are very noisy), the fixed period determination algorithm specifies, and nevertheless gives the value P_tThus by having a fixed period value P_tSo that method 10 can continue to be performed in subsequent iterations. Here, if there are not enough valid sums of the periodic samples, the period P is counted in step 500C_tTaken to be equal to the fixed period P determined in the previous iteration_t-1. It is reminded here that, in the initialization phase, there is no satisfactory first fixed period value P_tP can be obtained by applying a theoretical model depending on the engine operating conditions_t。

The measurement-based initialization phase, voltage signal S obtained by applying method 10 to the terminals of a transonic wheel sensor, is shown in fig. 6 and 7_VThe results obtained above. On the first axis of these figures, the voltage S as a function of time is shown_VMarks the pair signalS_VRestart threshold U of Schmitt trigger performing squaring_RAnd reset threshold U_A. On the second axis, a fixed period P obtained by applying the method 10 over the same time range is represented_tAnd another period value P_tiThe period value P_tiBy applying the same signal S_VObtained by applying a simple algorithm averaging periodic samples. Finally, the third axis shows the binary initialization signal U calculated according to the above conditions in the same time range_IThe value of (c).

Value P_tiCorresponds to a period value that does not distinguish between valid and invalid samples (corresponding to the criteria specified in step 200) and is not applied according to the signal S_VObtained by summing successive periodic samples measured on the square wave signal. For calculating the period P_tiComprises the same initialization criteria as those involved in the initialization test, so that the signal U can be managed_I. P may be initialized if the mutual dispersion of a certain number of periodic samples does not exceed the dispersion threshold D_tiAnd (4) calculating. Thus, in FIG. 6, the AND signal U_IThe time abscissa corresponding to the value 0 (initialization phase) of (a) and (b) and (c)_tiThe values of (a) are not relevant. On the other hand, in the initialization phase, method 10 uses the theoretical model P described above_thTo provide P_tValues of (A), these P_tCan be seen on the second axis of fig. 6.

Once the measurement is correctly initialized and the signal U_IIncreasing to a value of 1, the value P is obtained by taking the arithmetic mean of all periodic samples contained in the buffer B_ti. At the same time, the value P is obtained according to method 10_tThus, if the algorithmically determined number is greater than the first threshold A₁Then only the average of the valid samples is taken and, according to the above conditions, the sum of consecutive samples is taken in the opposite case. In step 200 of method 10, all samples of buffer B are determined to be valid for most of the time range shown in fig. 6, e.g., the time range in which the initialization measurement is made and before the range labeled ZAnd period P_tCorresponding to the arithmetic mean of all these samples. Therefore, in this time frame, the period P_tIs equal to the period P_ti。

On the other hand, it can be observed that, in the first axis of FIG. 6, the signal S is in the range Z_VIs larger than the previously obtained signal S_VThe value of (c). This represents an alternating current type of spurious noise which is added on top of the desired signal when measuring the voltage across the terminals of the tone wheel sensor. Within this range Z, many periodic samples among the samples taken and stored in the circular buffer B are "divided" and correspond to the signal S_VSpurious switching of the associated square wave signal. In fig. 7, it can be observed that the first and second axes of fig. 6 are plotted in a smaller time range of the time range Z. P_tiIs always obtained by taking the arithmetic mean of the same number of periodic samples, although many of these samples are "split" and correspond to a duration much smaller than the period of the desired signal. Thus, when the signal S is_VVery noisy, P_tiCan reach 1.0.10^-5S over the entire measuring time range, wherein the signal S is present_VWhen not very noisy or noiseless, P_tiA value of about 2.0.10^-4And s. On the other hand, an algorithm for calculating the fixed period (which finds the coherence value of the desired signal period from the "segmented" period stored in the circular buffer B), P, is used_tThe value hardly decreases in the range Z because it does not fall below 1.7.10^-4And s. Thus, the period P_tIs hardly affected by parasitic ac noise in the range Z.

The method 10 described above therefore makes it unnecessary to identify those signals corresponding to spurious noise in the switching of the square-wave signal and those corresponding to the desired signal for measuring the speed of rotation N1, the effect of the spurious noise of the alternating type on the period of passage of the measuring tone teeth being greatly reduced. Thereby making the measurement of the tone wheel speed and the measurement of the speed N1 more reliable. As mentioned above, this solution is easy to implement in an application system of the engine digital interface, which is much more efficient than a simple averaging over a constant number of periodic samples. Furthermore, this solution is advantageous with respect to the known solutions applying a low-pass type filter, since it does not affect the accuracy and sensitivity of the measurement and does not cause a phase shift of the signal acquired by the terminals of the transonic wheel sensor.