WO2020128274A1

WO2020128274A1 - Autonomous station for monitoring and analysing a maritime environment

Info

Publication number: WO2020128274A1
Application number: PCT/FR2019/053085
Authority: WO
Inventors: Eric Blin
Original assignee: Suez Groupe
Priority date: 2018-12-17
Filing date: 2019-12-16
Publication date: 2020-06-25
Also published as: FR3090178A1; FR3090129A1; FR3090129B1

Abstract

The invention relates to a marine monitoring and analysis station (100) comprising a floating base and an electric storage battery (110); characterised in that the marine monitoring and analysis station (100) comprises: - an acoustic acquisition member (140) for an underwater environment; and - an acoustic analysis device (130) for analysing the acoustic signal coming from the acoustic acquisition member (140), including computation means (131) for detecting by frequency analysis of said acoustic signal at least one pulse signal emitted by a living organism and/or a transient signal emitted by a living organism or coming from a boat.

Description

Title of the invention: Autonomous station for monitoring and analyzing a maritime environment The present invention relates to an autonomous station for monitoring and analyzing a maritime environment.

In particular, in maritime areas subject to significant human activity, such as coastal areas, coastal development, wind power installation areas or even wastewater discharge areas, monitoring of the maritime environment is necessary to analyze the impact of these activities on the marine environment.

To this end, water quality monitoring is generally carried out by direct observation and by the analysis of water samples taken. However, these analysis solutions, which are generally carried out in the laboratory after sampling the water, have relatively high costs and impose a minimum analysis time which does not allow sufficiently rapid detection of harmful events such as pollution. of water, a change in the turbidity of the water or an increase in disturbing sound frequencies of the underwater environment.

Autonomous buoys, such as Argos buoys, allow autonomous measurements and are intended to transmit environmental data relating to the marine environment in which they are installed. However, this type of beacon is generally expensive, bulky, and is not suitable for monitoring water quality in a coastal or near-coastal environment.

Also, there is the need for an autonomous maritime monitoring and analysis station, suitable for measuring the marine environment in an area with high human activity, such as a near-coastal, coastal, wind area or even a waste water discharge area.

To this end, a maritime monitoring and analysis station is proposed comprising a floating base and an electric storage battery; characterized in that the maritime monitoring and analysis station comprises: an acoustic acquisition unit for an underwater environment;

an acoustic analysis device for analyzing the acoustic signal coming from the acoustic acquisition device comprising calculation means for detecting by frequency analysis of said acoustic signal at least one impulse signal emitted by a living organism and / or a transient signal emitted by a living organism or from a boat.

Thus one can obtain a monitoring and analysis station capable of detecting various signals emitted by living organisms and by boats, in particular passages from boats, by acoustic analysis of the underwater environment; this in particular makes it possible to detect possible problems with the quality of the underwater environment and the sound pressures which can influence its fauna.

Such a station makes it possible to monitor the underwater environment, and more particularly the near-coastal or shallow underwater environment, in a relatively simple manner, by implementing the detection of relevant signals representative of the underwater fauna by the only acoustic study of the environment studied.

Advantageously and in a nonlimiting manner, the means for calculating the acoustic analysis device are adapted to determine, when a plurality of impulse signals is detected, whether at least part of said plurality of impulse signals forms a rhythmic sequence or not rhythmic. Thus, one can discriminate the different types of impulse signals from an underwater environment, for example discriminate the impulse signals emitted by the benthos, not rhythmic, from those emitted by the rhythmic odontocetes.

Advantageously and in a nonlimiting manner, said calculation means are adapted to carry out detection by frequency analysis in a plurality of predetermined frequency bands comprising at least one of the frequency bands from a first frequency band [200 Hz - 400 Hz] suitable for the detection of transient signals corresponding to a first category of fish, for example ophidions and corbs, a second frequency band [600 Hz - 1200 Hz] suitable for the detection of transient signals corresponding to a second category of fish, for example fish herbaria, and a third frequency band [100hz - 4000 Hz] for the detection of transient signals corresponding to the passage of boats. These frequency bands make it possible to target in a relatively precise manner the frequencies suitable for detecting signals making it possible to assess the evolution of the underwater environment, in particular in a shallow or near-coastal environment.

In particular, the combined detection of transient signals in these three frequency bands and the detection of rhythmic and non-rhythmic impulse signals makes it possible to obtain a relatively precise analysis of the underwater environment heard. In particular, a weakening of the emission of non-rhythmic impulse signals, corresponding in particular to clicks emitted by the benthos, makes it possible to determine a modification of the sedimentary medium or close to the bottom, such as pollution of the medium, a modification of the currents of background, a change in acidity, or a change in temperature of the medium. A weakening of the emission of rhythmic impulse signals, which can correspond to clicks of odontocetes, can also be significant of a modification of the turbidity of water, underwater currents, pollution or a sound disturbance of the environment for example by vibrations. The reduction in the detection of transient signals in the first and / or second frequency band also makes it possible to detect a change in the quality of the water in the bottom areas. In addition, the detection of the passage of boats, by detection of transient signals in the third frequency band, makes it possible to correlate the other data with a possible modification of the occupation of the maritime space. These acquired data thus make it possible to detect, in particular in the near coastal area, the impact of the direct environment, such as work in progress, wind installations, water discharge zones, on the underwater environment, quickly and relatively simply.

Advantageously and in a nonlimiting manner, the maritime monitoring and analysis station comprises a main calculation unit adapted to calculate analysis data from said at least one impulse signal and / or transient signal detected by said calculation means of said acoustic analysis device. The main calculation unit is, according to the main embodiment distinct from the calculation means of the acoustic analysis device. However, according to alternative embodiments, the main calculation unit may include the means for calculating the acoustic analysis device.

Thus, analysis data can be obtained locally, calculated in particular as a function of the acoustic data detected by the acoustic analysis device, and also by cross-checking with other signals coming from general modifications concerning the quality of the water: pollution , change in turbidity, change in currents, acidity

Advantageously without limitation, the main calculation unit comprises means for storing the calculated analysis data, and comprises means for evaluating the temporal evolution of said analysis data so as to determine the evolution of the medium submarine analyzed. Thus, the evolution of the underwater environment can be estimated by the evolution over time of the emission of the various descriptors of detected signals.

Advantageously and in a nonlimiting manner, the maritime monitoring and analysis station further comprises a set of acquisition means adapted to acquire data for measurements of said underwater environment, said main computing unit controlling the acquisition of data. of measures to said set of acquisition means. Thus the acoustic analysis data can be cross-checked and enriched with analysis data coming from acquisition means, such as physical sensors and / or digital data acquisition means for receiving data acquired from servers and / or databases, remote or on-board, so that monitoring and analysis is even more precise.

Advantageously and in a nonlimiting manner, the maritime monitoring and analysis station further comprises a device for acquiring meteorological data suitable for transmitting meteorological data to the main calculation unit. Thus the acoustic analysis data can be cross-checked and enriched with meteorological analysis data, obtained by means of acquiring meteorological data, for example by an on-board meteorological station or by remote access to a meteorological data server, so the monitoring and analysis is even more precise and allows more environmental parameters to be taken into account.

Advantageously and in a nonlimiting manner, the maritime monitoring and analysis station comprises radiocommunication means for communicating with a remote terminal; said main calculation member being adapted to transmit to the remote terminal the analysis data calculated by the radiocommunication means, so as to allow monitoring of the evolution of the underwater environment.

Thus, the on-board management of the calculation of data analyzes associated with the radiocommunication means makes it possible to reduce energy consumption by limiting the amount of data sent by the radiocommunication means to the remote terminal, while allowing remote acquisition of the analysis data.

According to an alternative implementation of the invention, the station can include wired communication means, for example for communicating with a remote coastal terminal, installed on a boat, on an off-shore station or on any other type of platform maritime.

Advantageously and in a nonlimiting manner, said main calculation member is adapted to transmit the measurement data in a common data frame with the analysis data calculated by the acoustic analysis device. Thus, it is possible to reduce the duration of data transmission to the remote terminal, which allows an optimization of the use of the on-board electric storage battery.

Advantageously and in a nonlimiting manner, the main calculation unit is adapted to control the activation of the radiocommunication means only during the duration of data transmissions to the remote terminal. Thus, it is possible to optimize the electrical use of the electric storage battery by reducing the consumption of the radiocommunication means by reducing their ignition time.

Advantageously and in a nonlimiting manner, the maritime monitoring and analysis station comprises geolocation means, the main calculation member being adapted to receive geolocation data from said geolocation means. So we can get data relatively precise positioning of the station, to take into account a possible drift, or to detect the problem area within the framework of a set of stations installed at sea.

Advantageously and in a nonlimiting manner, the maritime monitoring and analysis station comprises one or more photovoltaic panels recharging the electric storage battery by capturing solar energy. Thus, we can ensure total autonomy at the station, by ensuring an energy recharge during sunshine.

According to an alternative embodiment, the station can be powered by a connection to an external energy source, for example powered by directly from a terrestrial electrical source, by a boat, by an off-shore station or any other type of maritime platform. . This external power supply can be associated with or replace all or part of the electric storage battery and solar panels.

The invention also relates to a set for monitoring and analysis of a maritime environment comprising a remote terminal and at least one maritime monitoring and analysis station as described above, capable of acquiring and analyzing data from the maritime environment and to transmit them to said remote terminal by radiocommunication means.

Other particularities and advantages of the invention will emerge on reading the description given below of a particular embodiment of the invention, given by way of indication but not limitation, with reference to the appended drawings in which:

[Fig. 1] is a flow diagram of a maritime monitoring and analysis station according to an embodiment of the invention;

[Fig. 2] is a detailed schematic view of an acoustic analysis device according to the embodiment of Figure 1;

[Fig. 3] is a schematic view of a monitoring assembly according to an embodiment of the invention;

[Fig. 4] is a flow diagram representing in the form of a method a set of logical instructions implemented by an acoustic analysis device of the station according to the embodiment of FIG. 1; [Fig. 5] is a flowchart representing an alternative implementation of the method according to FIG. 4;

[Fig. 6] is a flow diagram of the steps of the first process of the method according to the embodiments of FIGS. 4 and 5;

[Fig. 7] is a schematic view of the steps of transforming the signal and of calculating a plurality of time series of percentage of occupation for a plurality of frequency bands of interest of the second process according to the embodiments of FIGS. 4 and 5 ;

[Fig. 8] is a schematic representation of the detection of signals of interest by exceeding the threshold of the second process according to the embodiments of FIGS. 4 and 5;

[Fig. 9] is a schematic representation of the characterization step of the events detected by the second process according to the embodiments of Figures 4 and 5; and

[Fig. 10] is a representation of a spectrogram and of graphical representations of the stages of the second process according to the embodiments of FIGS. 4 and 5.

According to a first embodiment of the invention, with reference to FIGS. 1 to 3, a monitoring and analysis station 100 of a maritime environment, to which reference will be made more simply by the term station 100, comprises means acquisition and acoustic treatment of the underwater environment, and means of measurement and analysis of the underwater and meteorological environment.

The monitoring and analysis station 100 is in this embodiment intended to be installed in a coastal or near-coastal area, in particular in a shallow area.

The station 100 comprises a floating base, not shown, waterproof, in which is installed a set of sensors 101, a meteorological measurement device 200 as well as an acoustic analysis device 130.

The station 100 is electrically powered by an electric storage battery 110. In the main embodiment of the invention, a set of photovoltaic panels 120 is installed on an upper external wall of the floating base.

The photovoltaic panels 120 make it possible to collect solar energy and recharge the electric storage battery when the sun is sufficient.

Thus, the electric storage battery 1 10 operates on the buffer battery principle, supplying energy on the one hand to the electronic components of the station 100 and on the other hand storing the energy received from the photovoltaic panels.

The station 100 is electrically dimensioned so as to be able to be electrically autonomous by daytime charging of the electric storage battery 110.

The station 100 is electronically structured around a main calculation unit 230 adapted to collect all of the measurement data supplied by the set of sensors 101, data supplied by the meteorological measurement device 200 as well as data supplied by the acoustic analysis device 130.

This main calculation unit 230 comprises in this embodiment a microprocessor, a random access memory and / or a storage memory, and means for receiving the data, such as a communication bus, for receiving the data coming from the set of sensors 101, the meteorological measurement device 200 as well as the acoustic analysis device 130.

The acoustic analysis device 130 of the underwater environment receives a signal from an acoustic acquisition device 140, here a submerged hydrophone 140, installed in external projection of the floating base and comprising a ceramic membrane.

The acoustic analysis device 130 receives a digital signal from the acoustic acquisition device 140.

However, if the hydrophone, or more generally the sound acquisition unit 140 provides only an analog signal, provision is made, in an alternative embodiment of the invention, for an analog conversion unit. digital, not shown, allowing to sample the sound recorded by the hydrophone in order to analyze it.

The acoustic analysis device 130 comprises a storage memory 132, for example a hard disk or a flash memory for storing the acoustic signal coming from the hydrophone, a random access memory 133, and a calculation member 131 for implementing a process for the acoustic analysis of a source acoustic signal from the acoustic acquisition device 140.

The computing member 131 is in a preferred embodiment of the invention a processor comprising at least two cores.

However, the invention is not limited to this type of processor and can, according to alternative embodiments, comprise a single core or be made up of several processors operating in parallel.

The acoustic analysis device 130 implements acoustic analysis method steps stored in the form of logic instructions in the storage memory 132 and / or loaded in the random access memory 133. Reference is made in the following description to this succession of logical instructions as being process steps.

These process steps form two independent processes 1, 2, which are implemented in this embodiment by two distinct cores of the processor 131.

The two processes 1, 2 form a process 6, 6 ’implemented by the acoustic analysis device 130 and in particular by the calculating member 131.

This method 6, 6 ’comprises stages of detection and categorization of signals emitted by marine animals, living aquatic organisms and boats. This method comprises in particular two distinct detection processes 1, 2:

a first detection process 1 of impulse signals, for example clicks emitted by the benthos, and clicks emitted by the odontocetes, but also any other impulse signal that can be emitted by a living underwater organism; and a second detection process 2 of transient signals, such as fish vocalizations, signals emitted by the passage of boats, or any other transient underwater signal.

In the present description, the term pulse signal or click means any short duration signal covering a relatively large frequency band, in particular any signal of duration less than 1 ms covering a frequency band ranging from 2000 Hz to 40,000 Hz.

The term transient signal, or vocalized signal, means any variable signal having a defined duration in a relatively targeted frequency band, for example a signal having a maximum energy less than 2 kHz and consisting of a succession of variations, also called pulses, of which the duration is generally between 10 ms and 100 ms per pulse, the signal can be of a duration of up to several seconds.

The two processes receive as input 60 the same sound source, also called the acoustic source signal, such as a recorded file, compressed or not, comprising the entire sound sequence to be analyzed. According to this embodiment, the sound sources analyzed are recordings of substantially fifteen consecutive minutes, but their duration is not limiting and can be for example between five minutes and one hour.

The longer a record, the more reliable the analyzed data, but the longer it takes to process. A fifteen-minute recording provides a relatively optimal compromise between analysis time and the relevance of the data analyzed.

According to an alternative implementation of the invention, the sound source can also be a source acoustic signal transmitted in real time, however the analysis requiring the detection in time of frequency events, a storage of the received signal is then necessary, for example over a period of between fifteen minutes and one hour.

The source acoustic signal is acquired for example by a hydrophone, or any other means of underwater sound capture.

This source acoustic signal can be acquired in the form of a digital signal by temporal sampling of the captured analog sound. Such a method of transforming an analog sound source into a signal Numerical is not detailed in the present description, however the sampling frequency must necessarily be greater than twice the highest frequency potentially detected, in accordance with Shannon's theorem.

These two processes 1, 2 are implemented in parallel, as represented in FIG. 1, according to the main embodiment of the invention, for example by a same processor by division of tasks, by two hearts of the same processor, or by two separate processors.

These two processes 1, 2 can, according to an alternative embodiment, with reference to FIG. 2, be carried out consecutively one after the other, without necessarily pre-established order, immediately or separated by a predetermined period of time. .

However, these two processes 1, 2 form a whole, the final result of which corresponds to the detection and categorization of two types of signals: impulse signals and transient signals, so as to allow the analysis of the underwater environment heard .

All the signals detected and categorized thus make it possible to analyze the evolution of an underwater environment, for example by detecting an overall sound evolution of the underwater environment.

The main objective of the invention is not, however, to discriminate precisely between each underwater species emitting a sound signal in a given environment. Although the person skilled in the art is able to associate between 600 and 700 species of fish with their vocalization frequency, all the vocalization signatures are not known or sufficient to ensure precise detection of the species in a sound stream. , especially since several species can coexist in the same frequency band.

The invention mainly aims to determine and more generally discriminate the types of signals present so as to be able to evaluate the general evolution of the underwater ecosystem heard. However, the invention can still be implemented in a similar way, by targeting narrow frequency bands specific to certain election bodies so as to study their evolution independently of the general ecosystem. According to the embodiment described as an example of implementation of the invention, without limitation, it is detected:

- clicks emitted by the benthos, in the form of a non-rhythmic impulse signal;

- clicks emitted by the odontocetes, in the form of a rhythmic impulse signal;

- fish vocalizations, in the form of a transient signal; and

- the passage of boats, in the form of a transient signal.

This thus makes it possible to obtain a relatively precise sound knowledge of the underwater environment heard, in particular in a near-coastal environment, or at shallow depth, for example at depths less than 100m.

In other words, this method makes it possible to transform a raw acoustic signal into a plurality of descriptors of the underwater environment, comprising:

- rhythmic impulse signals and the percentage of sound coverage of these signals over a period of analysis;

- non-rhythmic impulse signals and their average number over an analysis period;

- transient signals associated with their number per unit of time and their sound power.

Thus, the underwater and marine environment studied can be analyzed relatively precisely by detecting the majority of identifiable sound sources, in a relatively simple and rapid process.

To this end, the first detection process 1, which aims to detect impulse signals comprises a first step of transformation 10 of the source acoustic signal in a frequency domain.

This transformation of the source acoustic signal is implemented by the application of a discrete local Fourier transform, better known by its English abbreviation Discrete-Time STFT, for Short-Term Fourier Transform.

The operation here is an operation of a discrete nature because the processed signal is a digital signal. The local Fourier transform, also called according to the uses Fourier transform with sliding window, short-term Fourier transform or more commonly transformed with sliding Fourier, allows obtaining a spectrogram representing the signal in a time-frequency plane . Such a spectrogram is defined as the square modulus of the short-term Fourier transform.

The spectrogram represents the energy of the signal in time, according to the segment analyzed, and in frequency, according to the Fourier coefficient.

Also, the step of calculating the spectrogram as a function of the short-term Fourier transform is also called the step of calculating the signal energy.

In this embodiment of the invention, a step of calculating the signal energy is implemented in a frequency band adapted to the detection of impulse signals coming from underwater living organisms, for example benthic clicks and / or clicks of odontocetes, here a frequency band ranging from 2000 Hz to 40,000 Hz. However, one can select one or more different frequency bands, depending on the species and categories of organisms to be detected; in particular, this step could be adapted to narrower frequency bands, so as to target certain organisms more precisely.

The signal transformed by the sliding Fourier transform is subdivided into a plurality of segments, in this embodiment 2000 segments on a signal of fifteen minutes.

However, the invention is not limited to this duration which can for example be chosen in a period of time between five minutes and one hour.

The sliding Fourier transform is parameterized with 2048 points, a recovery rate of 50% and an apodization per Kaiser window with side lobe attenuation with a value of 180 dB.

Then, a step of determining a signal detection threshold l, for each instant t, in the frequency band is carried out, by analysis of the energy levels, and a noise foot N, corresponding to a noise level N in the frequency band of the studied signal. Within the meaning of the invention, the noise N present in the source signal is an Gaussian additive noise, ie a totally random signal, which therefore corresponds to white noise, of normal distribution in the time domain, with zero mean. Consequently, the noise N has a constant power spectral density corresponding to the ratio of the power of the noise by the corresponding bandwidth. However, this power spectral density is updated over time, according to a frequency chosen by a person skilled in the art.

This step of determining the signal detection threshold l can be carried out according to several methods, such as a predetermination of the values before the implementation of the method, or by way of example a statistical method by segmentation of the time-frequency plane, using the reallocation vector technique, or statistical methods such as binary tests of hypotheses based on the theory of detection, in particular via Bayesian or Neyman-Pearson approaches.

In this embodiment of the invention, the noise foot N of the signal and the detection threshold l are estimated according to the time-frequency plane segmentation method disclosed in the scientific publication Dadouchi, F., Gervaise, C., loana, C., Huillery, J., & Mars, JI (2013). Automated segmentation of linear time-frequency representations of marine mammal sounds. The Journal of Acoustical Society of America, 134 (3), pp. 2546-2555, the content of which must be considered to be part of this request.

This method allows a determination by thresholding of the best noise foot N pair of the signal / detection threshold l, the noise foot N corresponding to different noise levels of a set of spectrograms, and the threshold l making it possible to maximize the detections of rhythmic and non-rhythmic clicks by optimally reducing the number of false positives, by applying a Neyman-Pierson test for a plurality of binary spectrograms defined for a plurality of noise levels. We then proceed to a selection of the parameters allowing to obtain the best ratio alerts / false alerts due to noise. The previous detection method comprises steps of defining a set of binary spectrograms corresponding to different noise values N, then for each binary spectrogram, detecting the number of alerts for a plurality of detection threshold values l, and selection of the best pairs of noise foot N / detection threshold values l, reducing the probability of false detections.

Once this pair of noise foot N and threshold l values has been obtained, the second order statistic s2 of the signal is calculated, which will be referred to more simply as being the statistic of the signal s2.

In fact, the source signal is considered in the present embodiment as a random signal corresponding to a time sequence of random variables.

These random variables are linked by a statistical link between successive samples. This relationship is frequently characterized by the joint probability density of the two corresponding random variables.

The sequence of functions of the joint probabilities of the pairs of random variables is what is called the second order statistic of the random signal.

This second order statistic of the random signal corresponds in this embodiment to the signal power spectral density, which is obtained by the Fourier transform of its autocorrelation function.

Also, obtaining the two-order statistics s2 of the signal is implemented by calculating the Fourier transform of the autocorrelation function of the signal analyzed segment by segment.

One then proceeds to the detection 13 of impulse signals by calculating the ratio of the order 2 statistics of the signal with the statistical estimate of the noise foot N by comparison with the detection threshold l.

When a plurality of interest pulses, also called interest signals or even event signals, are then detected, a time-rhythm transform of the series of instants of interest pulses detected in the signal to reveal the existence of a rhythm.

This time-rhythm transform comprises in this embodiment the calculation of the autocorrelation of the source signal. Indeed, as indicated above, we seek to discriminate two types of clicks: non-rhythmic clicks and rhythmic clicks.

Non-rhythmic clicks are for example impulse signals generated by the benthos, repeated at a frequency generally between 50 and 200 pulses per second, with no remarkable rhythm.

Rhythmic clicks, impulse signals for example generated by odontocetes, have a repetition frequency of a click or a sequence of clicks that can range from 5 to 800 clicks per second, repeated at regular intervals.

It is for example known that a sequence of clicks emitted by a dolphin can be repeated every 2 ms, while a sequence of clicks emitted by a sperm whale can be repeated every 2 seconds.

A rhythm is defined by the regularity of the repetition of the click or the sequence of clicks over time. For example, ten successive repetitions of the same click or the same sequence of clicks at regular intervals form a rhythmic sequence.

You can define a margin of error in the duration of the repetition interval, for example 5% of the duration of the interval, to qualify the regularity of the repetition.

In this exemplary embodiment, it is considered that a rhythm, for example for an odontocetus rhythm, is detected when a rhythm is detected ranging from 5 clicks per second to 50 clicks per second.

Conversely, non-rhythmic clicks are generally issued irregularly, with a countdown of around 50 clicks per second.

Also, we can obtain the average number per second of non-rhythmic clicks, and the percentage of time covered by rhythmic clicks over the time range studied.

The second process makes it possible to detect and distinguish the vocalizations of fish and the passage of boats.

This second process 2 comprises a first step 20 of transformation of the source signal in the time-frequency domain. This first step 20 is carried out according to a calculation of a discrete local Fourier transform, as described for the first process 1, making it possible to obtain a spectrogram.

The discrete local Fourier transform, as for the first process 1, is parameterized with 2048 points, a recovery rate of 50% and an apodization by Kaiser window with attenuation of the side lobes of a value of 180 dB

However, for a detection of fish the transformation of step 20 will be configured as a function of a frequency band dependent on the categories of fish that one seeks to detect, for example a bandwidth of [200 Hz, 400Flz] to detect a first category of fish, such as ophidions and corbs, a bandwidth of [600 Hz, 1200 Hz] to detect a second category of fish such as herbarium fish, and with a gross spectrogram over 15 seconds.

The bandwidth (s) are defined to group together sets of categories of fish to be detected in the underwater environment studied.

Regarding the detection of boat passages, we could for example select a bandwidth of [100 Hz, 4000 Hz], with an averaged spectrogram to give 2000 time segments of 15 seconds. The sliding Fourier transform being parameterized with 2048 points, a recovery rate of 50% and an apodization by Kaiser window with attenuation of the side lobes of a value of 180 dB. This bandwidth of [100 Hz, 4000 Hz] is generally wide enough to detect a majority of boats.

The spectrogram obtained thus makes it possible to obtain in a time-frequency plane the frequency distribution of the signal energy.

A second step 21 is then implemented of calculating a plurality of time series of percentage of occupancy for a plurality of frequency bands of interest.

In this exemplary embodiment, the second step 21 is implemented for respectively two frequency bands for detecting fish: a first frequency band [200 Hz - 400 Hz], in particular suitable for the detection of ophidions and corbs, a second frequency band [600 Hz - 1200 Hz] especially adapted for the detection of fish from seagrass beds, and a third frequency band [100hz - 4000 Hz] for the detection of boat passages.

This second step 21 therefore corresponds to the calculation of the spectral energy density of the signal in each specific frequency band, as a function of time, and to the calculation of the percentage of the spectral energy densities in each band of interest with respect to the total signal energy for each instant.

However, the invention is not limited to these particular frequency bands and the method can be implemented for any frequency band of interest relating to the emission of vocalizations of other species of fish or other bands of frequencies which may relate to the passage of ships.

Thus, during the second step 21, a percentage of signal coverage is obtained for each acquisition instant for each frequency band of fish detection, also called the frequency band of interest.

Regarding the interest band of boat detection, the second step of the process is implemented over a period of 15 minutes, by averaging the spectrogram over 2000 segments. In other words, the spectrogram is reduced to 2000 analyzed segments, each segment corresponding to the frequency average of the energies of the spectrogram over periods of substantially 0.45 sec per segment. Indeed, the passage of boats is a phenomenon more spread out in time, and the detection analysis is then carried out over a longer duration and requires less temporal precision.

Conversely, the calculation of the coverage rate with regard to the interest bands for detecting fish is carried out by gross analysis of the spectrogram at 15-second intervals.

Then, for each frequency band of interest, the signals 23 of interest are detected by exceeding the threshold of the percentage of signal coverage in each fish detection frequency band. In other words, when the percentage coverage of the signal in a frequency band of interest exceeds a threshold value at a given time, a signal of interest is then detected.

The threshold values 25 can vary depending on the organisms or objects detected.

In particular, in this exemplary embodiment, the threshold is 100% for the ophidions and the corbs.

The threshold 25 is 80% for fish from seagrass beds.

The threshold is 50% for the passage of boats.

For each threshold crossing, we proceed to a step of categorizing the event 24, by calculating the duration of the event, which corresponds to the time elapsing between the rising edge 240 of exceeding the threshold 25, and the edge descending 241 during which the percentage of coverage falls below the threshold 25.

We then implement a step of estimating the signal energy during the duration of the event, which makes it possible to characterize the event.

Finally, we count over a period of analysis time, here between 15 minutes and one hour, the number of events detected in each frequency band of interest.

Thus, the second process 2 makes it possible to count the number of vocalizations of fish, in this embodiment by distinguishing on the one hand the corbs and ophidions, on the other hand the fish from the herbaria. This second process 2 also makes it possible to detect and estimate the level of sound energy associated with the passage of boats.

In this way, it is possible to obtain a unified method making it possible to detect a plurality of living organisms as well as the passage of boats from a single source of sound acquisition, in a relatively simple manner and relatively optimized in computing time.

In particular, the following analysis data, also called descriptors, are obtained from a source acoustic signal, corresponding to an underwater sound recording, for example a 15 minute recording:

- the average number of benthic clicks per second; - the percentage of time covered by clicks of odontocetes;

- the number of boat passages per unit of time as well as the associated sound energy;

- the number of fish vocalizations per unit of time in each frequency band analyzed as well as the associated sound power in each frequency band.

From this determined analysis data, a step of calculating the evolution of the underwater environment which is the subject of the recording of the source acoustic signal is carried out, according to the different analysis data obtained by the two processes 1 , 2, in relation to previously acquired analysis data, so as to evaluate the evolution of the underwater environment heard.

In the main embodiment, method 6, 6 ’is carried out a plurality of times. At each iteration of the process, the analysis data are stored in a storage memory.

The calculation step also carries out a comparative analysis of the evolution of the analysis data compared to previous data so as to determine an evolution of the underwater environment.

In particular, a reduction or spacing in time of the emissions of non-rhythmic clicks, such as benthos clicks, make it possible to determine a modification of the sedimentary medium or close to the bottom, such as pollution of the medium, a modification of the currents of background, a change in acidity, or a change in temperature of the medium.

A weakening of the emission of clicks of odontocetes can also be significant of a modification of the turbidity of water, underwater currents, pollution or even a noise disturbance of the environment for example by vibrations.

The reduction in the detection of fish from seagrass beds also makes it possible to detect changes in water quality in the bottom areas.

The detection of the passage of boats makes it possible to correlate the other data with a possible modification of the occupation of the maritime space.

These data can also be correlated with each other to determine a common impact, taking into account in particular a modification of the known environment, such as works, a discharge area or an increase in maritime traffic.

These acquired data thus make it possible to detect, in particular in the near coastal area, the impact of the direct environment, such as work in progress, the presence of wind installations, water discharge zones, on the environment under -sea, quickly and relatively simply.

The measurement data can, according to alternative embodiments of the invention, be compared to fixed, predetermined data or to digital models providing standard data for a given underwater area.

In particular, these data allow the station 100 to obtain information on the evolution of the turbidity of the water, in a relatively simple manner and without it being necessary to use complex and cumbersome methods such as only by acoustic Doppler profiler or by nephelometry methods. The presence of fish and odontocetes having a causal link with the turbidity of the water.

The station 100 can also be adapted to calculate, with reference to FIG. 10, the exposure sound level, also called SEL 27 (for Sound Exposure Level), corresponding to the logarithm of the sound energy received over a duration of exposure here equal to 1 second, the unit of which is dB / second.

The station 100 can also calculate the sound pressure level, also called SPL (for Sound Pressure Level) expressed in dB.

The SEL and SPL data descriptors are part of the data descriptors that the station 100 can transmit to the remote terminal 300 as described below.

The station 100 further includes a set of sensors 101.

The set of sensors 101 comprises a current measurement probe 15, more generally called a current meter 150. In particular a Doppler effect current meter 150 is implemented in the main embodiment, which is a relatively reliable current meter implementation in the maritime environment. However, it is also possible to use, according to alternative embodiments of the invention, a propeller current meter, more generally known by the name of hydrometric reel. The monitoring and analysis station is also equipped with a gyroscope 180 making it possible to detect the orientation of the currents relatively simply, by detecting the orientation of the station 100.

The set of sensors 101 also includes a probe for measuring the water level 160, for example a sonar 160.

The sensor assembly 101 also comprises at least one of a salinity measurement probe 170, an oxygen dissolved in water probe 190, for example a galvanic micro-sensor probe, a temperature probe of the water.

A dissolved oxygen sensor 190 makes it possible in particular to detect water pollution when the dissolved oxygen value decreases. This information can be correlated with acoustic analysis data, in particular the evolution of the presence of fish and odontocetes, to attest to significant water pollution.

We can also plan to install a pH probe. However, the evolution of pH in a marine environment generally varies over quantities of the order of 1 / 100th of a unit. However, the known sensors for such a maritime monitoring and analysis station 100 do not make it possible to obtain sufficient sensitivity. Also, the evolution of the pH of the water, and its alkalinity, are not carried out by probe, but are derived from the analysis of the quality of the water carried out by the acoustic analysis.

The station 100 also includes a meteorological measurement device 200 comprising a barometer, a thermometer for measuring the outside temperature and an anemometer.

A geolocation unit 210 is also installed in the maritime monitoring and analysis station 100. This geolocation unit 210 is in the main embodiment of the invention a GPS 210. However, the invention is not limited to this only mode of geolocation.

In the main embodiment of the invention, where the station 100 is intended to be installed in near-coastal areas, a radiocommunication device 220, also called radiofrequency device 220, is on board to communicate with a remote terminal 300. However, the invention is not limited to communication with a remote terminal 300. In particular an alternative embodiment of the invention implements all of the calculations in station 100. In other words all of the calculations and operations performed by the remote terminal 300 in the main embodiment are carried out directly in the station 100 in this alternative implementation of the invention.

In this embodiment, the radiocommunication device 220 is a device suitable for communicating via cellular networks, for example on a GSM or GPRS network. However, it is possible to provide, in alternative embodiments of the invention, other means of communication such as satellite communication devices.

The radiocommunication device 220 allows two-way communication, called full-duplex, with the remote terminal 300. In other words, the station 100 is capable of transmitting data and of receiving information from the remote terminal 300, for example control instructions aimed at to order where to plan measurements and analyzes to be carried out.

The station 100 includes a main calculation unit 230 intended to collect all of the measurement data supplied by the set of sensors 101, the measurements of the meteorological measurement device 200 as well as the acoustic detection data calculated by the measurement device. acoustic analysis 130.

This main calculation unit 230 comprises in this embodiment a microprocessor, comprising a cache memory, a random access memory, a storage memory and means for receiving data coming from the set of sensors 101, from the measurement device meteorological 200 as well as the acoustic analysis device 130.

According to an embodiment of the particular invention, the main calculation member 230 forms a single member with the calculation means 131 of the acoustic analysis device 130, in particular it is a single processor 131, 230 .

The main calculation unit 230 is also connected to a real-time clock, more frequently called under its English abbreviation Real-Time Clock, which supplies the main calculation unit with the current date and time. The calculation unit 230 controls the start and duration of the data acquisition to the set of sensors 101, to the meteorological measurement device 200 as well as to the acoustic analysis device 130, via the acquisition device. acoustic 140.

This start and duration of acquisition data can either be obtained in the form of a remote command sent from the remote terminal during a period of availability of the radio communication means 220 or according to a predetermined schedule, stored in memory, for example according to an interval of predetermined regular time calculated according to data from the real time clock.

According to the main embodiment of the invention, the calculating member 230 is on the one hand remotely controllable, in particular during the period of listening to the predetermined radio communication means 220 or during the sending of frames data from the station 100 to the remote terminal 300, and the computation unit 230 is preprogrammed and can be reprogrammed remotely by the remote terminal 300, so as to proceed automatically, at a predetermined instant, at a predefined regular interval , or during a predefined period of time for the acquisition of measurement and analysis data.

These acquisition data are then transmitted to the main calculation unit 230 by each device from among the set of sensors 101, the meteorological measurement device 200 and the acoustic analysis device 130.

The main calculation unit 230 generates from the information received, a data frame, time-stamped according to the date and time provided by the real-time clock, and comprising a predetermined data set.

This frame is then transmitted to the remote terminal 300 via the radio communication means 220.

The frame includes all of the data analyzed. Also, the data transmitted are not raw data, the data extraction having been carried out by the main calculation unit 230 and the acoustic analysis device 130. The main calculation unit 230 is suitable for controlling the automatic switching on and off of the radiocommunication means 220 outside of the data sending periods in order to save the energy consumption of the station 100.

According to a particular implementation of the invention, the main calculation unit 230 can be configured to send particular analysis data when an analyzed data leaves an associated range of normal values. In other words, when an abnormal event or data is detected, the main calculation unit 230 is then configured to transmit the data to the remote terminal 300.

According to a second embodiment of the invention, the station 100 is augmented with additional sensors. In particular, station 100 includes a fluorometric sensor, for measuring chlorophyll a by fluorescence. Indeed the measurement of chlorophyll a makes it possible to detect an algal bloom, more generally known under the name of algal bloom, or in English algal bloom. Such algal bloom is in a known way correlated with the planktonic mass present in the underwater environment studied. This therefore makes it possible to obtain additional relevant information on the underwater environment analyzed.

According to a third embodiment, in which the station 100 already includes the sensors and analysis means of the first or of the second embodiment, bacteriological analysis means are also installed, so as to refine any pollution detections some water.

According to a fourth embodiment of the invention, in which the station 100 is adapted to be installed on the high seas, or in zones of medium to great depths, ie generally above 40 meters in depth, certain sensors can be secured to cables connected to the floating base, so as to allow deep immersion of these sensors.

In particular, the hydrophone 140 can be submerged, secured to the probe body or offset via a cable, to a depth making it possible, for example, to pick up sounds close to the bottom. Also, the study of currents is of interest both at the surface and at the bottom, so the station 100 can include a plurality of current meters, in particular immersed at different depths, via cables, so as to measure different values of currents.

All the sensors in the sensor set 101 can be multiplied and removed from the floating base to be immersed at different depths.

To this end, the station 100 can also include a plurality of hydrophones, and the acoustic analysis means are then adapted to implement the acoustic analysis steps for each acquired acoustic signal, so as to analyze the evolution of the environment. submarine at different depths. The data acquired in the same time interval at different depths are then cross-checked to refine the analysis data.

To ensure increased durability of the station 100, automatic maintenance and seawater protection devices can be installed on the station 100. For example, copper grids can be installed for their algicidal action and / or brushes , for example flexible blades or a set of flexible fibers, coming to remove sensors 101, by an automated mechanical action of the brush, the foreign bodies being able to be deposited there, such as algae, shrimps, mussels etc., known phenomenon under the English term of Fouling.

In addition, it is possible to install, for certain sensors 101, a constitution of titanium or any other suitable material, making it possible to provide reinforced protection to seawater.

Such a monitoring and analysis station according to one of the preceding embodiments, can be adapted according to the use made of it. In particular, depending on the analysis environment, for example a wind field, coastal site monitoring, impact monitoring on a wastewater discharge area, the on-board sensors can be added or removed as necessary and in such a way as to optimize energy consumption.

Claims

[Claim 1] Maritime monitoring and analysis station (100) comprising a floating base and an electric storage battery (1 10); characterized in that the maritime monitoring and analysis station

(100) includes:

an acoustic acquisition unit (140) of an underwater environment; and an acoustic analysis device (130) for analyzing the acoustic signal coming from the acoustic acquisition device (140) comprising calculation means (131) for detecting by frequency analysis of said acoustic signal at least one impulse signal emitted by a living organism and / or a transient signal emitted by a living organism or coming from a boat.

[Claim 2] Maritime monitoring and analysis station (100) according to claim 1, characterized in that the calculation means (131) of the acoustic analysis device (130) are adapted to determine, when a plurality of pulse signals is detected, if at least a portion of said plurality of pulse signals forms a rhythmic or non-rhythmic sequence.

[Claim 3] Maritime monitoring and analysis station (100) according to claim 1 or 2, characterized in that said calculation means are adapted to carry out detection by frequency analysis in a plurality of predetermined frequency bands comprising at least one of the frequency bands from a first frequency band [200 Hz - 400 Hz] adapted for the detection of transient signals corresponding to a first category of fish, a second frequency band [600 Hz - 1200 Hz] adapted for the detection of transient signals corresponding to a second category of fish, and a third frequency band [100hz - 4000 Hz] for the detection of transient signals corresponding to the passage of boats. [Claim 4] Maritime monitoring and analysis station (100) according to any one of claims 1 to 3, characterized in that it further comprises a main calculation member (230) suitable for calculating data from analysis from said at least one impulse signal and / or transient signal detected by said calculation means (131) of said acoustic analysis device (130).

[Claim 5] Maritime monitoring and analysis station (100) according to claim 4, characterized in that the main calculation unit (230) comprises means for storing the calculated analysis data, and comprises means for evaluation of the temporal evolution of said analysis data so as to determine the evolution of the analyzed underwater environment.

[Claim 6] Maritime monitoring and analysis station (100) according to claim 4 or 5, characterized in that it further comprises a set of acquisition means (101) suitable for acquiring measurement data of said medium submarine, said main computing unit (230) controlling the acquisition of measurement data from said set of acquisition means (101).

[Claim 7] Maritime monitoring and analysis station (100) according to any one of claims 4 to 6, characterized in that it further comprises a meteorological data acquisition device (200) suitable for transmitting meteorological data to the main computing unit (230).

[Claim 8] Maritime monitoring and analysis station (100) according to any one of claims 4 to 7, characterized in that it comprises radio communication means (220) for communicating with a remote terminal (300); said main calculation unit (230) being adapted to transmit to the remote terminal (300) the analysis data calculated by the radiocommunication means (220), so as to allow monitoring of the evolution of the underwater environment. [Claim 9] Maritime monitoring and analysis station (100) according to claim 8, characterized in that said main calculation unit (230) is adapted to transmit the measurement data in a data frame common with the data d analyzes calculated by the acoustic analysis device (130).

[Claim 10] Maritime monitoring and analysis station (100) according to claim 8 or 9, characterized in that the main calculation unit (230) is adapted to control the activation of the radiocommunication means only for the duration data transmission to the remote terminal (300).

[Claim 1 1] Maritime monitoring and analysis station (100) according to any one of claims 1 to 10, characterized in that it comprises geolocation means (210), the main calculation unit (230 ) being adapted to receive geolocation data from said geolocation means. [Claim 12] Maritime monitoring and analysis station (100) according to any one of claims 1 to 11, characterized in that it comprises one or more photovoltaic panels (120) recharging the electric storage battery (1 10) by collecting solar energy. [Claim 13] Set for monitoring and analysis of a maritime environment comprising a remote terminal (300) and at least one monitoring and analysis station (100) according to any one of claims 8 to 12, capable acquiring and analyzing data from the maritime environment and transmitting it to said remote terminal (300) by radio communication means (220).