GB2290641A - Method of evaluating signals - Google Patents

Description

2290641 METHOD OF EVALUATING SIGNALS The present invention relates to a

method of evaluating signals, particularly signals requiring accurate evaluation for component or system fault detection, such as in motor vehicle diagnostic equipment.

In the evaluation of signals which consist of a sequence of measurement values, problems arise when the signal analysis takes place by way of a microprocessor having a computing capacity which is not as great as desired. Moreover, since the signals are changed by events of different kinds, an accurate analysis of the signals and consequent diagnosis or recognition of actions or faults is very involved and usually requires a large computing capacity.

The evaluation of signal courses for diagnosis is known in connection with, for example, the monitoring of a rotational speed transmitter in a motor vehicle from German (Federal Republic) laid open specification No. 40 26 232. In this evaluation, the course of the vehicle on-board mains voltage is evaluated by a control device.

The voltage decreases sharply after actuation of the vehicle engine starter and increases again in characteristic manner after some time.

When such a voltage course is recognised by the control device, it is concluded therefrom that starter actuation has taken place. The output signal of the rotational speed transmitter, which records the rotational speed of the engine crankshaft, should then allow recognition that the crankshaft is rotating. If the output signal of the transmitter does not change, a defect in the transmitter or a line interruption must be present.

This known diagnostic equipment, which evaluates the course of the onboard mains voltage for monitoring of the transmitter, has the disadvantage that the voltage collapses not only in consequence of actuation of the starter, but also when other loads are switched on, especially heavy loads. Moreover, the voltage level is dependent on temperature and is subject to a number of disturbances.

According to the present invention there is provided a method of evaluating signals with a sequence of measurement values, comprising the steps of convoluting a first signal, which is to be evaluated, with a second and substantially rectangular or step-shaped signal, comparing the signal obtained from the convolution with a threshold, and recognising, on the threshold being exceeded, the occurrence of an event which has triggered an influence on the first signal.

Alternatively, the obtained signal can be compared with pattern functions and, on agreement thereof with corresponding patterns, an associated event which has triggered an influence on the first signal is recognised.

A method exemplifying the present invention may allow a very reliable signal evaluation to be carried out independently of interference, while at the same time the computing capacity required for the evaluation may be kept within reasonable limits. The signal to be evaluated, which can consist of a sequence of measurement values obtained by analog-to-digital conversion, is convoluted with a second signal and the signal arising after the convolution is 3 - compared with a threshold, whereby an event is recognised when the threshold is exceeded. Due to the convolution with a second signal which is rectangular or has a step, there is obtained a signal shape which is particularly easy to evaluate.

It is particularly advantageous that recognition of a number of occurring events is possible, so that monitoring of a wide variety of components is possible. The method is especially suitable for monitoring loads or sensors in motor vehicles.

If the course of the on-board mains voltage in a motor vehicle 1 is evaluated as a function of time as the first signal, a variety of events which influence the voltage can be recognised after convolution with the second signal. For example, it can be recognised whether a lamp that has been switched on is actually functioning and leads to a measurable voltage collapse or whether, in the case of switching on of lights, both lamps are alight or only one.

Through the evaluation of signal courses and subject to consideration of different plausibility conditions, a diagnostic system can be provided which operates reliably even without convolution with a further signal.

Examples of the method of the present invention will now be more particularly described with reference to the accompanying drawings, in which Fig. 1 is a schematic block circuit diagram of a battery voltage monitoring system with signal evaluating means; Fig. 2 is a diagram illustrating steps in performing a signal evaluation by a method exemplifying the invention; Fig. 3 is a diagram showing a first signal, which is to be evaluated by such a method; Fig. 4 is a diagram showing a second signal to be used in convolution of the first signal in accordance with that method; and Fig. 5 is a diagram showing the relationship between the first signal and the resulting convoluted signal.

Referring now to the drawings, there is shown in Fig. 1 a simple form of monitoring system incorporating signal evaluating means which can be used in connection with a method exemplifying the invention. In the illustrated system, evaluation is carried out by a microcontroller or computer 1, which evaluates the voltage of a battery 2. Connected to the battery 2 is a load 3, and in parallel therewith, by way of a switch, a lamp 4. The voltage across the battery is ascertained by a voltage meter 6 and communicated to the microcontroller 1 by way of an anal og-to-digital converter 7 at an input E1. Further signals for evaluation, for example measurement values from sensors 8, can be fed to the microcontroller at an input E2 thereof. The microcontroller 1 controls a display 9 by way of an output A, and also the switch 5.

The system of Fig. 1 is valid as a general illustration of signal evaluation applied to voltage monitoring, and an example of an actual evaluating method will now be described in more detail by way of Fig. 2. Fig. 2 shows a diagnostic method for use in a vehicle, in which block 10 symbolises the process and thus, in this case, represents the vehicle in which the actions to be monitored take place, for example switching on of loads. Input magnitudes 11, for example switching signals or signals indicative of driver wishes, act on the block 10. In addition, faults 12 and unknown input magnitudes 13, for example the engine rotational speed course or a switching operation, act on the block 10.

Resultant signal courses 14, for example vehicle battery voltage, engine temperature and vehicle wheel speed, are measured by sensors or other suitable means and applied to a block 15. Through appropriate signal processing, the engine rotational speed can be derived from the battery voltage, for example by determining the spacing between ignition pulses which lead to voltage collapses, the spacing being inversely proportional to the rotational speed. Thus, the battery voltage, the engine temperature, the wheel speed, the engine speed and also data about switching events can be collected in the one block 15, in which an extraction of events takes place.

Data concerning individual events are available at the output of the block 15. These data can be in the form of characteristic values or a signal course. The individual events are evaluated in a status observer 16, in which a comparison of the events takes place with remainder generation.

Of the status observer 16 itself, blocks 17, 18 and 19 are illustrated, wherein the bock 17 symbolises a vehicle model and the block 18 represents a diagnostic test in which it is checked whether a remainder is plausible. Finally, the individual events as well as the ascertained remainder, thus the deviation between measured and stored parameters, are fed to the block 19. The status observer 16 - 6 also comprises a microprocessor or microcontroller, in which computations take place, in particular convolutions of signals.

Still further. possibly required magnitudes, such as wheel rotational speed (from block 15) or the input magnitudes 11 can be applied to the vehicle model 17. The model stores different parameters which describe the actual state, for example the aforementioned collapse of the vehicle mains voltage on actuation of a lamp (flasher, high beam and so forth). When it is recognised, for example, that the flasher was actuated, the characteristic values, in this case the voltage collapse, are computed from the signal. The diagnostic system now compares these two characteristic values and, according to any deviation therebetween, makes a decision about the functional capability of the lamp. Generally, the diagnostic system can be described as a component which monitors the vehicle model for plausibility. It examines each new event to ascertain whether it is plausible and for this purpose monitors the long-term trend of the parameters to ascertain whether these, too, are plausible or whether they indicate a fault.

Input magnitudes of different kinds can be evaluated in the diagnostic system illustrated in Fig. 2. Plausibility tests take place to allow recognition of the occurrence of faults or of correct functioning of equipment. Which of the input magnitudes is evaluated at a particular time depends on different requirements. Important input magnitudes are checked frequently and less important ones only occasionally.

Fig. 3 shows a signal S1, which occurs more frequently in normal operation of the motor vehicle. In particular, the signal S1 represents the course of the vehicle battery voltage UB as a function of time t when a flasher of the vehicle is briefly actuated. The numbers MW = 0 to 200 represent measurement values. The signal course S1 is characteristic and reproducible. On switching on of the lamp, the lamp filament is still cold and thus has a very low resistance, for which reason a large current flows. The large current causes a strong voltage collapse. However, since the filament heats up rapidly, the resistance quickly rises and the voltage levels off to a constant value. After the lamp is switched off, the voltage returns to the usual value.

The voltage course illustrated in Fig. 3 is idealised, i.e. no other load is switched on during this time and no disturbances have arisen within this time. Subject to these presumptions, an evaluation of the voltage course would be possible to allow recognition of whether or not the lamp functions correctly, in that the functioning of the lamp can be concluded if the voltage falls below a limit value UG1. However, since the voltage course is usually subject to disturbances and the battery voltage, is as a matter of principle, dependent on temperature, monitoring of functional capability by this means is problematic. 20 This difficulty can be overcome, in a method exemplifying the invention, by convoluting the voltage course or signal S1, as illustrated in Fig. 3, with a step-shaped signal S2 of zero mean value. A new convoluted signal, which can be evaluated in a particularly reliable manner, then results. 25 The signal S2, i.e. a step of zero mean value, is illustrated in Fig. 4. Fig. 5 shows the signal S1, i.e. battery voltage course, and, in dashed lines, the convoluted signal S3. Whilst the unloaded battery voltage fluctuates in dependence on temperature in a range between 13 and 13.5 volts, the convoluted signal S3 (Illustrated with voltage US3 in Fig. 5) is usually equal to zero and merely the characteristic maximum or minimum values are clearly different from zero. This relationship is completely independent of whether the battery voltage in the unloaded state is at a higher level or in the unloaded state or at low temperature at a lower level.

In view of these relationships, a threshold UG2 can be fixed, which, when exceeded, allows a definite signal to be recognised. In the case of the described example, it is recognised as soon as the convoluted signal exceeds the threshold that a lamp has been switched on. By changing the height of the threshold value, it becomes possible to recognise whether, when high beam is switched on, merely one lamp is operating or both lamps are operating. Such a recognition is possible because operation of both lamps causes a doubling of the voltage collapse by comparison with the operation of only one lamp.

The convolution of two signals is a procedure known in electrical technology and is described in, for example, "DTV-Lexikon der PhysiC, Volume 3, Issue of August 1970, page 84. There is understood, by the convolution product of two signals or two functions, G 1 (t) and G 2(t)l the integral:

+00 G (t) GI(T) G2(t-T) dT -00 The symbol for the convolution product is a star. Thus, G(t) G, (t) G 2(t). For evaluation, a series of Fourier transforms is usually required. If, however, the convolution is performed with a - 9 step-shaped signal S2 as described in Fig. 3, a convoluted signal S3 is obtained with simple evaluation. This signal is largely independent of direct voltage components and relatively insensitive to high-frequency interference. The signal S2 illustrated in Fig. 3 5 initially has a signal height H1 over a length Ll. After this length Ll, the signal S2 has a height H2 and retains this height for a length L2. Such a signal can be designated as a step of zero mean value, wherein ML1 = HM2 in order for the condition of zero mean value to be obtained.

The convolution can be realised by, for example, two sliding mean values which are carried out one behind the other and then added up. Through use of the described step signal S2, however, a simple recursive formulation is also possible, which can easily be worked out and does not require any significant computing effort. A formula for the convolution is Y n = Y n-1 + M(X n - X n-Ll) - H2( Xn-L1 - Xn-Ll-L2) wherein Ynl is the convolution result of the preceding convolution, % is the actual measurement value, Xn-L1 is the measurement value for nLl, and so forth.

If the heights H1 and H2 are powers of two, the convolution with the step signal can be carried out with the aid of additions, loading commands and shift commands. These commands can be easily carried out with a microcontroller. An event is recognised when the signal resulting from the convolution exceeds the value of the threshold.

By appropriate dimensioning of the lengths Ll and L2 it is ensured that the characteristic frequencies of the signals to be recognised are suppressed as little as possible. The filter formed by the convolution has a band-pass degree characteristic with a very low lower limit frequency. The upper limit frequency is dependent on the parameters L1 and L2.

The described evaluation of signals by way of convolution with a step signal having a zero mean value permits a variety of fault recognitions, or recognitions that a functional capability is present, to be performed. In a simple system such as that of Fig. 1, for example, it can be recognised by the microcontroller when the switch 5 is actuated for the ignition circuit of the lamp 4, in which case a voltage collapse can be expected. If the convoluted signal then exhibits the expected course, it is recognised that the switched-on lamp is functioning correctly. If, however,, the convoluted signal fails to exceed the fixed threshold value UG2, the microcontroller 1 recognises that the lamp is not functioning correctly and accordingly activates the display 9.

In a more complex system, data about the signal course, for example the course of the battery voltage, are continuously fed to the microcontroller 1. New convoluted signals are constantly formed by running convolution of each measurement value of the signal to be evaluated, wherein clear associations result from the signal course arising from the convolution by comparison with previously stored pattern signal courses. Thus, for example, it is possible to recognise from the convoluted signal that switching on of the high beam lamps has taken place, wherein the lamps conduct current simultaneously. By comparison with a pattern function, the supply of a control signal of the actuated switch can be dispensed with, whereby a further possibility of fault is suppressed in the diagnosis. After the recognition of fault, further checks can be initiated.

Apart from monitoring the course of the battery voltage, further plausibility checks can be carried out in respect of processes which take place inevitably. These plausibility checks can take place for convoluted signals or for unconvoluted signals, with a high reliability being attainable in the case of convoluted signals. An improvement can be achieved if several signal courses are analysed.

A further possibility of diagnostic monitoring can concern a vehicle radiator. When the vehicle is at rest, as recognised from a tachosignal, and the engine is idling, conclusions concerning the state of the radiator can be drawn from the speed of rise in the temperature. Yet another possibility consists of monitoring the rate of change in the temperature. If the temperature changes significantly during travel without immediately obvious reason, the driver can be warned before damage occurs.

Diagnostic monitoring can also be carried out in respect of a central locking system of a vehicle. In most such systems, the duration of the locking operation changes when one of the parts of the system does not function correctly. This can be detected by monitoring the course of battery voltage.

Claims (13)

1 A method of evaluating signals with a sequence of measurement values, comprising the steps of convoluting a first signal, which is to be evaluated, with a second and substantially rectangular or step- shaped signal, comparing the signal obtained from the convolution with a threshold, and recognising, on the threshold being exceeded, the occurrence of an event which has triggered an influence on the first signal.

2. A method of evaluating signals with a sequence of measurement values, comprising the steps of convoluting a first signal, which is to be evaluated, with a second and substantially rectangular or stepshaped signal, comparing the signal obtained from the convolution with pattern functions, and recognising, on agreement of the pattern functions with corresponding patterns, an associated event which has triggered an influence on the first signal.

3. A method according to claim 1 or claim 2, wherein the second signal is a step function with a mean value of zero and having a first height over a first length and a second height over a second length.

4. A method according to claim 3, wherein the first and second lengths are so selected as to cause minimum attenuation of the frequency components, which are to be evaluated, of the first signal.

5. A method according to any one of the preceding claims, wherein the first signal is the voltage of a vehicle battery or of a vehicle mains system and said event is the connection to the battery or the system of an electrical load causing a temporary voltage collapse.

6. A method according to any one of the preceding claims, comprising the step of carrying out evaluation of the first signal by means of a microcontroller or a computer and initiating a control action, an indicating action or a control and indicating action in dependence on the evaluation.

7. A method according to claim 3, wherein the step of convoluting is carried out in accordance with the formula Y n = Y n-1 + W(Xn_ X n-Ll) "(Xn-Ll-XnLl-L2)l wherein H1 and L1 are the first height and length, respectively, of the second signal step function, H2 and L2 are the second height and length, respectively, of the second signal step function, Y n-1 is the result of a preceding convolution of the first signal, Xn is the actual measurement value, X n-Ll is the measurement value for n-11 and X n-Ll-L2 is the measurement value for n-Ll-L2.

8. A method according to claim 7, wherein the heights H1 and H2 are powers of 2 and the convolution is carried out with the assistance of additions, loading commands and shift commands.

9. A method of evaluating signals with a sequence of measurement values, comprising the steps of analysing at least one signal course, carrying out comparisons with events to be expected, carrying out parameter monitoring and recognising a fault when a signal course 5 does not lead to an expected event.

10. A method according to claim 9, wherein the step of parameter monitoring comprises analysing the speed of increase in the temperature of vehicle coolant in response to recognition of idling of the vehicle engine with the vehicle at rest.

11. A method according to claim 9, wherein the step of parameter monitoring comprises analysing the rate of change in the temperature of vehicle coolant and recognising a fault in the case of a rate of change exceeding a predetermined threshold rate.

12. A method according to claim 9, wherein the step of parameter monitoring comprises monitoring the duration of operation of a central locking system of a vehicle by analysing the voltage of the vehicle battery and recognising a fault in the case of a duration exceeding a predetermined threshold duration.

13. A method according to claim 1 and substantially as hereinbefore described with reference to the accompanying drawings.