DEVICE AND METHOD FOR MEASURING AND TRACKING THE QUANTITY OR CONCENTRATION OF A COMPOUND IN A FLUID

0Associated
Cases 
0Associated
Defendants 
0Accused
Products 
0Forward
Citations 
0
Petitions 
1
Assignment
First Claim
1. A device for measuring and tracking over time the quantity or concentration of a component in a fluid, comprising:
 a sensor capable of measuring a quantity or concentration of the component in the fluid and providing a quantitative signal for tracking this quantity or concentration over time,a signalprocessing module comprising a lowpass filter of the quantitative tracking signal,an output interface for providing the filtered quantitative tracking signal,characterized in that the signalprocessing module comprises;
an estimator of a value of instantaneous trend of variation of the quantitative tracking signal in a predetermined sliding time window, andmeans for adjusting over time a high cutoff frequency of the lowpass filter according to the estimated value of instantaneous trend of variation.
1 Assignment
0 Petitions
Accused Products
Abstract
A device for measuring and tracking over time the quantity or concentration of a component in a fluid comprises: a sensor capable of measuring a quantity or concentration of the component in the fluid and providing a quantitative signal for tracking this quantity or concentration over time; a signalprocessing module comprising a lowpass filter of the quantitative tracking signal; and an output interface for providing the filtered quantitative tracking signal. The signalprocessing module comprises an estimator of a value of instantaneous trend of variation of the quantitative tracking signal in a predetermined sliding time window. Also provided is means for adjusting over time a high cutoff frequency of the lowpass filter according to the estimated value of instantaneous trend of variation.
0 Citations
No References
No References
10 Claims
 1. A device for measuring and tracking over time the quantity or concentration of a component in a fluid, comprising:
a sensor capable of measuring a quantity or concentration of the component in the fluid and providing a quantitative signal for tracking this quantity or concentration over time, a signalprocessing module comprising a lowpass filter of the quantitative tracking signal, an output interface for providing the filtered quantitative tracking signal, characterized in that the signalprocessing module comprises; an estimator of a value of instantaneous trend of variation of the quantitative tracking signal in a predetermined sliding time window, and means for adjusting over time a high cutoff frequency of the lowpass filter according to the estimated value of instantaneous trend of variation.  View Dependent Claims (2, 3, 4, 5, 6, 7, 8)
 9. A method for measuring and tracking over time the quantity or concentration of a component in a fluid, comprising the following steps:
measuring a quantity or concentration of the component in the fluid and providing a quantitative signal for tracking this quantity or concentration over time, using a sensor, processing the quantitative tracking signal using a lowpass filter, providing the filtered quantitative signal at the output, characterised in that the processing of the quantitative tracking signal comprises; an estimation over time of an instantaneous trend of variation of the quantitative signal in a predetermined sliding time window, and the adjustment over time of a high cutoff frequency of the lowpass filter according to the estimated instantaneous trend of variation.
 10. A computer program downloadable from a communication network and/or recorded on a medium readable by computer and/or executable by a processor, comprising instructions for the execution of the following steps:
receiving a quantitative digital signal for tracking a quantity or concentration over time of a component in a fluid, processing the quantitative tracking signal by lowpass filtering, characterised in that the lowpass filtering is carried out using instructions for the execution of the following steps; estimating over time an instantaneous trend of variation of the quantitative signal in a predetermined sliding time window, and adjusting over time a high cutoff frequency of the lowpass filtering according to the estimated instantaneous trend of variation.
1 Specification
The present invention relates to a device for measuring and tracking over time the quantity or concentration of a component in a fluid. It also relates to a corresponding method and computer program.
The invention applies more particularly to a device comprising:
 a sensor capable of measuring a quantity or concentration of the component in the fluid and providing a quantitative signal for tracking this quantity or concentration over time,
 a signalprocessing module comprising a lowpass filter of the quantitative tracking signal,
 an output interface for providing the filtered quantitative tracking signal.
The industrial uses are multiple, for the detection of gaseous, liquid or solid components in any gaseous or liquid fluids. The sensors that can be used are also multiple and depend on the intended uses. A nonlimiting example is the search for certain gaseous species in the air, such as identified pollutants and/or greenhouse gases, using optical detection methods based on the spectral absorption properties of the various species that can compose a gas and on the BeerLambert law. The sensor can in this case be of the NDIR type (from “NonDispersive InfraRed”), that is to say having a nondispersive infrared emitter and, in general, a thermopile detector.
A device of this type is for example described in the patent application US 2003/0058439 A1 or in the patent application WO 2007/064370 A2. Other devices using the same optical method but further having optimized production costs and bulk are also marketed by the applicant with MEMS technology (from “MicroElectroMechanical Systems”). They can be integrated into portable electronic systems such as tablet computers, mobile phones, cameras or other. They can also be integrated into stationary systems in home automation, industrial detection or analysis of air quality. The signals that they provide can be advantageously used for a display of information or for a triggering of an alert in the case of an identified danger.
But for a better use of the signals provided by this type of device, said signals only rarely come directly from the sensor. They generally undergo a processing comprising at least a slight lowpass filtering aimed at removing certain noises or artefacts from the signals. This filtering further causes a smoothing facilitating the reading of the information or the interpretation of the detection. In return, it introduces a latency between the estimated quantity or concentration of component and the actual quantity or concentration at each instant. This latency can raise an issue in certain uses in which a rapid increase or decrease in a component, judged dangerous or vital, in a fluid requires precise detection and increased reactivity (triggering of an alarm or of a plan of actions). A compromise, which is always unsatisfactory, must thus be found between the effectiveness and the latency of the lowpass filtering.
It can thus be desired to provide a device which allows to at least partially overcome the aforementioned compromise.
A device is therefore proposed for measuring and tracking over time the quantity or concentration of a component in a fluid, comprising:
 a sensor capable of measuring a quantity or concentration of the component in the fluid and providing a quantitative signal for tracking this quantity or concentration over time,
 a signalprocessing module comprising a lowpass filter of the quantitative tracking signal,
 an output interface for providing the filtered quantitative tracking signal. wherein the signalprocessing module comprises:
 an estimator of a value of instantaneous trend of variation of the quantitative tracking signal in a predetermined sliding time window, and
 means for adjusting over time a high cutoff frequency of the lowpass filter according to the estimated value of instantaneous trend of variation.
By acting on the high cutoff frequency of the lowpass filtering, action is generally taken directly on the latency: the lower the high cutoff frequency, the smoother and easier to interpret the quantitative tracking signal but the higher the latency introduced; on the contrary, the higher the high cutoff frequency, the lower the latency but the more noisy and difficult to interpret the quantitative tracking signal. Thus, the invention allows to dynamically adapt this high cutoff frequency according to an estimation of variation of the quantitative tracking signal in a predetermined sliding time window which can be much shorter than the latency of a lowpass filter, the parameters of which result from an unsatisfactory compromise as mentioned above. By doing so, the lowpass filtering is adapted in real time to the variability of the signal and the compromise becomes dynamic and thus satisfactory.
Moreover, it is noted that the notion of lowpass filter includes that of bandpass filter. Indeed, a bandpass filter must be considered to be a particular case of a lowpass filter, i.e. a lowpass filter further having a low cutoff frequency lower than its high cutoff frequency.
Optionally, the estimator is designed to estimate the value of instantaneous trend of variation of the quantitative tracking signal by providing a value of slope of this signal in the predetermined sliding time window, in particular via linear regression. This particularly simple estimation can be carried out in very short times.
Also optionally, the adjustment means are designed in such a way as to:
 reduce the high cutoff frequency of the lowpass filter when the absolute value of the estimated value of instantaneous trend of variation decreases, and
 increase the high cutoff frequency of the lowpass filter when the absolute value of the estimated value of instantaneous trend of variation increases.
Also optionally, the lowpass filter is designed to carry out, after temporal sampling of the quantitative tracking signal, a digital filtering by exponentially weighted moving average according to the following time recurrence relation:
where y_{1}, . . . , y_{i}, . . . are successive time samples of the quantitative tracking signal,
Also optionally, the adjustment means are configured using fuzzy logic in such a way as to:
 distinguish N states of instantaneous trend of variation, N≥2, each of these
N states being associated with a predetermined corresponding value of high cutoff frequency of the lowpass filter and with a membership function with values in the interval [0; 1] defined in a range of possible values of instantaneous trend of variation, and
 adjust the high cutoff frequency of the lowpass filter as a sum of the N predetermined values of high cutoff frequency respectively weighted by N degrees of membership of the estimated value of instantaneous trend of variation in each of the N states of instantaneous trend of variation, these degrees of membership being calculated using the N membership functions.
The configuration using fuzzy logic provides an adjustment that is reactive but without rupture of the lowpass filtering.
Also optionally, N≥3 and:
 a first stable state is associated with a high cutoff frequency linked to a value of the exponential weighting coefficient α between 0.9 and 1,
 a second state of slow variation is associated with a high cutoff frequency linked to a value of the exponential weighting coefficient α between 0.7 and 0.9,
 a third state of fast variation is associated with a high cutoff frequency linked to a value of the exponential weighting coefficient α between 0.1 and 0.3.
Also optionally, each membership function is a Gaussian or piecewise linear function. These functions are simple to implement using fuzzy logic.
Also optionally, the sensor is a gas sensor with a nondispersive infrared emitter and a thermopile detector. In this case, the invention allows a use in the detection of gas in a gaseous medium.
A method is also proposed for measuring and tracking over time the quantity or concentration of a component in a fluid, comprising the following steps:
 measuring a quantity or concentration of the component in the fluid and providing a quantitative signal for tracking this quantity or concentration over time, using a sensor,
 processing the quantitative tracking signal using a lowpass filter,
 providing the filtered quantitative signal at the output,
wherein the processing of the quantitative tracking signal comprises:  an estimation over time of an instantaneous trend of variation of the quantitative signal in a predetermined sliding time window, and
 the adjustment over time of a high cutoff frequency of the lowpass filter according to the estimated instantaneous trend of variation.
A computer program is also proposed, downloadable from a communication network and/or recorded on a medium readable by computer and/or executable by a processor, comprising instructions for the execution of the following steps:
 receiving a quantitative digital signal for tracking a quantity or concentration over time of a component in a fluid,
 processing the quantitative tracking signal by lowpass filtering,
the lowpass filtering being carried out using instructions for the execution of the following steps:  estimating over time an instantaneous trend of variation of the quantitative signal in a predetermined sliding time window, and
 adjusting over time a high cutoff frequency of the lowpass filtering according to the estimated instantaneous trend of variation.
The invention will be better understood with the help of the following description, given only as an example and made with reference to the appended drawings in which:
The device illustrated in
The sensor 10 is capable of measuring a quantity or concentration of the component C in the fluid F and providing a quantitative signal for tracking this quantity or concentration over time. According to the example precisely illustrated in
The signalprocessing module 12 receives the quantitative tracking signal y_{1}, . . . , y_{i}, . . . via an input interface 28. It further comprises a lowpass filter 30 and an estimator of instantaneous variation 32 to which it transmits this signal y_{1}, . . . , y_{i}, . . . . The estimator 32 is more precisely designed to calculate a value V of instantaneous trend of variation of the quantitative tracking signal in a sliding time window having a predetermined length T. Advantageously, this length T is much less than the latency that the lowpass filter 30 can cause. Moreover, according to the general principles of the invention, the module 12 also comprises means 34 for adjusting over time a high cutoff frequency f_{C }of the lowpass filter 30 according to the value of instantaneous trend of variation returned by the estimator 32. Finally, it comprises an output interface 36 for providing the filtered quantitative tracking signal. This filtered signal is noted as
In a preferred embodiment, the estimator of instantaneous variation 32 is designed to estimate the value V of instantaneous trend of variation of the quantitative tracking signal , y_{1}, . . . , y_{i}, . . . by providing a value of slope of this signal in the predetermined sliding time window, in particular via linear regression. Such a linear regression is relevant if it can be supposed that the quantitative tracking signal is approximately linear in the time window considered, which is often the case in short time intervals. By noting for example as x_{iT}, . . . , x_{i }the T sampling times of the sliding time window at a time x_{i }at which it is desired to estimate the value V, the linear regression involves expressing the corresponding tracking signal y_{iT}, . . . , y_{i }in the following form:
and determining the values and that minimise the zeromean residual errors ϵ_{iT}, . . . , ϵ_{i}.
By noting:
the value
which is the leastsquares solution by minimisation of the mean quadratic error is estimated in the following manner: Â=[X^{T}X]^{−1}.X^{T}.Y, where X^{T }is the transpose of the matrix X.
As a result, the value V of instantaneous trend of variation can be chosen as equal to the absolute value of the slope . As indicated by the above calculations and in accordance with the notion of sliding time window, this value V can be revised upon each reception of a new sample of quantitative tracking signal.
Also in a preferred embodiment, the adjustment means 34 are designed to:
 reduce the high cutoff frequency f_{C }of the lowpass filter 30 when the absolute value of the value V of instantaneous trend of variation calculated by the estimator 32 decreases, and
 increase the high cutoff frequency f_{C }of the lowpass filter 30 when the absolute value of the value V of instantaneous trend of variation calculated by the estimator 32 increases.
In this case, a direct relation between V and f_{C }can be predefined in the adjustment means, for example in the form of an increasing function, in order to adjust the high cutoff frequency f_{C }according to V. Since the value V can be revised upon each reception of a new sample of quantitative tracking signal, the high cutoff frequency f_{C }can also be adjusted at the same rate.
Also in a preferred embodiment, the lowpass filter 30 is digital and designed to carry out on the sampled quantitative tracking signal y_{1}, . . . , y_{n}, . . . a digital filtering by exponentially weighted moving average according to the following time recurrence relation:
where α is an exponentialweighting coefficient of the digital filtering by moving average between 0 and 1 and mathematically related to the high cutoff frequency f_{C }of this digital filtering. It can for example be demonstrated on the basis of the above time recurrence relation that this mathematical relation takes the following form for an estimation of the high cutoff frequency at −3 dB:
where F_{S }is the sampling frequency, cos^{−1 }is the inverse of the cosine function and max (;) is the function that returns the maximum between two values.
This mathematical relation results in particular from the teaching of the work by Rick Lyons, entitled “Understanding Digital Signal Processing”, 3^{rd }edition, Prentice Hall Publishing, 2011, pages 613614. It can be simplified in the following manner when f_{C }remains small compared to F_{S}, in particular as long as f_{C}≤0.1 F_{S}:
In this case, the direct relation between V and f_{C }can be predefined in total equivalence by a direct relation between V and α.
In a preferred embodiment, this direct relation is configured using fuzzy logic in the adjustment means 34 in the following manner:
 N discrete states of instantaneous trend of variation are defined and distinguished, N≥2, each of these N states being associated with a predetermined corresponding value f_{C}(n) of high cutoff frequency f_{C }of the lowpass filter 30, or in an equivalent manner with a predetermined corresponding value α(n) of the weighting coefficient α, and with a membership function F_{MS,n }with values in the interval [0; 1] defined in a range of possible values for V (for example included in [0; +∞[), and
 adjust the high cutoff frequency f_{C }of the lowpass filter 30, or in an equivalent manner the weighting coefficient α, as a sum of the N predetermined values of high cutoff frequency, or in an equivalent manner as a sum of the N predetermined values of the weighting coefficient, respectively weighted by
N degrees of membership of the estimated value V in each of the N states of instantaneous trend of variation, these degrees of membership being calculated using the N membership functions.
Each membership function is for example a Gaussian or piecewise linear function. Any other family of membership functions well known in fuzzy logic is also possible and can be adapted according to the needs of the intended use and the context.
The elements 30, 32, 34 of the signalprocessing module 12, as illustrated in
These functional modules thus comprise a plurality of computer programs or a plurality of functions of the same computer program, these programs or functions being able to be grouped together according to any possible combination into one or more pieces of software. They could also be at least partly microprogrammed or microwired into dedicated integrated circuits. Thus, alternatively, each computer device implementing one or more of the functional modules described above could be replaced by an electronic device composed only of digital circuits (without a computer program) for carrying out the same actions.
The module 14 for using data provided by the module 12 receives the filtered signal
It is noted that the sensor 10, the signalprocessing module 12 and the module 14 for use of data can be structurally separated. Thus, the device illustrated in
The operation of the device of
During a step 100 executed continuously by the sensor 10, the thermopile detector 18 provides a continuous analogue signal of measurement of light absorption by the component C of light radiation emitted by the infrared emitter 16 and transmitted by the gas F.
During a step 102 executed continuously by the converter 26 of the sensor 10, the continuous analogue signal provided by the thermopile detector 18 by execution of the step 100 is digitally converted in order to provide the successive time samples of the quantitative tracking signal y_{1}, . . . , y_{i}, . . . .
During a step 104 executed at each instant of the time sampling by the estimator 32, a value V of instantaneous trend of variation of the quantitative tracking signal y_{1}, . . . , y_{i}, . . . in a sliding time window having a predetermined length N is calculated.
During a step 106 executed at each instant of the time sampling by the adjustment means 34, the high cutoff frequency f_{C }of the lowpass filter 30, or in an equivalent manner the aforementioned weighting coefficient α in the case of digital filtering by exponentially weighted moving average, is adjusted according to the value V estimated in step 104.
During a step 108 executed at each instant of the time sampling by the lowpass filter 30 which applies the high cutoff frequency f_{C}, or in an equivalent manner the aforementioned weighting coefficient α in the case of digital filtering by exponentially weighted moving average, the quantitative tracking signal y_{1}, . . . , y_{i}, . . . is filtered and transformed into
This filtered quantitative tracking signal
A simple and concrete example of calculation of the aforementioned weighting coefficient α according to the value V estimated at each instant of time sampling is illustrated in
 a first stable state is associated with a high cutoff frequency linked to a value α(1) of the exponential weighting coefficient α between 0.9 and 1, for example α(1)=0.99;
 a second state of slow variation is associated with a high cutoff frequency linked to a value α(2) of the exponential weighting coefficient α between 0.7 and 0.9, for example α(2)=0.8;
 a third state of fast variation is associated with a high cutoff frequency linked to a value α(3) of the exponential weighting coefficient α between 0.1 and 0.3, for example α(3)=0.2.
The stable state is associated with a first membership function F_{MS,1 }with values in the interval [0; 1] defined in the range of values [0; +∞[. Fora simple illustration, this first membership function is piecewise linear in the example of
The slow state of variation is associated with a second membership function F_{MS,2 }with values in the interval [0; 1] defined in the range of values [0; +∞[. For a simple illustration, this second membership function is also piecewise linear. It first of all continuously takes the value 0 in the first interval, then increases linearly towards 1 in the second interval, then continuously takes the value 1 in a third interval, then linearly decreases towards 0 in a fourth interval, then continuously takes the value 0 in the rest of the range of values.
The fast state of variation is associated with a third membership function F_{MS,3 }with values in the interval [0; 1] defined in the range of values [0; +∞[. For a simple illustration, this third membership function is also piecewise linear. It first of all continuously takes the value 0 in the first, second and third intervals, then increases linearly towards 1 in the fourth interval, then continuously takes the value 1 in the rest of the range of values.
The weighting coefficient α is thus determined from the value V estimated at each instant of time sampling by the following relation:
α=F_{MS,1}[V]. α(1)+F_{MS,2}[V]. α(2)+F_{MS,3}[V]. α(3),
or, for example in the precise illustration of
α=0.8×0.99+0.2×0.8+0×0.2=0.952.
It is clear that a device for measuring and tracking like the one described above allows to measure and track, in a denoised manner and without bothersome latency, a quantity or concentration of a component C in a fluid F.
Moreover, it is noted that the invention is not limited to the embodiment described above in reference to
Thus for example, even though the lowpass filtering was proposed above directly on the signal coming from the thermopile detector 18, after sampling and digitisation, it could have also been proposed at another location in the processing chain, in particular after conversion according to the BeerLambert law. Nevertheless, this alternative is less advantageous because the exponential expression of the BeerLambert law leads to a significant amplification of the noised variations before lowpass filtering.
It will be clear more generally to a person skilled in the art that various modifications can be made to the embodiment described above, in light of the teaching that has just been disclosed to the person skilled in the art. In particular, even though the estimation of instantaneous variation of the quantitative tracking signal was advantageously recommended by linear regression because of the simplicity of this method, other methods known to a person skilled in the art for estimating a trend of instantaneous variation of a signal can be applied. Likewise, even though digital filtering by exponentially weighted moving average was recommended for the operation of the lowpass filter 30, other lowpass filters with an adjustable cutoff frequency, in particular other filters with infinite impulse response, are possible. Likewise, multiple direct relations between V and f_{C }can be defined by a person skilled in the art, even if the configuration using fuzzy logic has particular advantages and is judicious.
In general, in the following claims, the terms used must not be interpreted as limiting the invention to the embodiment disclosed in the present description, but must be interpreted to include therein all the equivalents that the claims aim to cover due to their wording and the providing of which is within the reach of a person skilled in the art by applying their general knowledge to the implementation of the teaching that has just been disclosed to them.