Vessel motion monitoring system and method for prediction and / or rapid detection of parametric roll

Abstract

A monitoring system for a parametric roll detection can trigger an alarm, potentially providing a phased alarm with various gradations of alarms. The monitoring system includes a device onboard the vessel that is capable of measuring the time signal of the vessel's roll. From the time signal of the vessel's roll, an algorithm is used to extract the vessel's natural roll period. A shift of the natural roll period that is significant can indicate an imminent parametric roll. The monitoring system can also make risk prediction based on the natural roll period. The risk predictions can warn the vessel operators of the risk of entering parametric roll, even when the natural roll period of the vessel is relatively constant. To make risk predictions, a risk estimator, in the form of a formula involving the sea state and the natural roll period of the vessel, is used.

Description

VESSEL MOTION MONITORING SYSTEM AND METHODFOR PREDICTION AND / OR RAPID DETECTION OF PARAMETRIC ROLLCROSS-REFERENCE TO RELATED APPLICATIONS

[0001] This application claims the benefit of priority to US provisional application serial nos. 63 / 450,324 filed on March 6, 2021, and 63 / 626,709 filed on January 30, 2024, which are incorporated herein by reference in their entireties for all and any purposes.BACKGROUND

[0002] This disclosure relates generally to vessel motion monitoring systems and methods. This disclosure relates more particularly to monitoring systems and methods for the prediction and / or the rapid detection of parametric roll of vessels.

[0003] There have been many incidents of significant loss of cargo which have primarily been caused by large roll angles that resulted from a phenomenon called "parametric roll."

[0004] American Bureau of Shipping, “Guide for the assessment of parametric roll resonance in the design of container carriers,” Rules and Guides, April 2019, 63 pages, pub. no. 133, Spring TX, USA, available at URL https: / / ww2.eagle.org / content / dam / eagle / rules-and-guides / current / design_and_analysis / guide-assessment-parametric-roll-resonance-design-containercarriers / parametric-roll-guide-aprl9.pdf describes the physics of this phenomenon, which is due to a confluence of wave conditions that interact with the vessel geometry to change the waterplane stiffness of the vessel such that the stability is significantly less than typical and, in some segments of the wave cycle, is characterized by negative roll stiffness.

[0005] Parametric roll can build up quickly to very large roll angle values that endanger the integrity of the cargo and, potentially, the vessel. Common knowledge in the industry is that the time from onset to extreme roll angles is so short that detection and warning are insufficient. The container shipping industry has therefore focused on prediction methods rather than detection / warning methods.

[0006] There are, however, actions that the captains can take to mitigate the parametric roll if it can be detected early enough. These mitigating actions include changing speed and direction of the vessel.SUMMARY

[0007] The reason that the time from onset to extreme rolls is perceived to be too short is that the industry uses maximum amplitude of roll angle to detect parametric roll. Maximum amplitude of roll angle builds up quickly: common knowledge indicates that two cycles are sufficient to go from normal amplitudes to extreme ones.

[0008] When parametric roll happens, some of the vessel's natural periods (e.g., natural roll period) can change significantly from their initial values, which are otherwise constant and are typically considered to be calculable based on standard naval architectural and the weight distribution of the cargo. It should be appreciated that the natural roll period depends on the vessel's structure and cargo.

[0009] Data from incidents indicate that the shift in natural roll period can often be detected well before there is a significant build-up of roll angle amplitude, thereby providing a sufficiently long time to react, provided that the natural roll period shift can be detected on short time signals.

[0010] The disclosure generally describes a monitoring system for a parametric roll detection. It is important to note that the parametric roll detection system can start trigger an alarm much before the other systems can, potentially providing a phased alarm with various gradations of alarm.

[0011] The monitoring system includes a device onboard the vessel that is capable of accurately measuring the time signal of the vessel’s motion, including the vessel’s roll. This device can comprise one or more Inertial Measurement Units (“IMUs”).

[0012] From the time signal of the vessel’s roll, a signal processing algorithm is used to extract the vessel’s natural roll period. A shift of the natural roll period that is significant can indicate the onset of parametric roll, which could mean an imminent rapid build-up of large roll angles. As used herein, the amount that qualifies for a significant shift can be determined using simulations and / or experimental data on any vessel and its cargo. Additionally, or alternatively, a shift that is considered significant can correspond to a roll angle having a sufficiently large amplitude. For example, changes in the natural roll period when the amplitude of the roll angle is five degrees peak-to-peak or less may be discounted as not significant, whereas changes in the natural roll period when the amplitude of the roll angle is ten degrees peak-to-peak or more may be analyzed more closely.

[0013] A spectrum (e g., obtained via Fourier Transform) is normally used for natural period detection, including the detection of the vessel’s natural roll period. Computing a spectrum relies on processing long time signals which would make rapid detection difficult. Thus, another signal processing algorithm is preferred.

[0014] In general, there may be three criteria for selecting a preferred signal processing algorithm: i) The preferred signal processing algorithm can accurately detect at least one cyclical period from a time series of measurements.

[0015] More particularly, the preferred signal processing algorithm may decompose the time series into one or a sum of signal(s), either purely sinusoidal or damped sinusoidal, each having a frequency and an amplitude that is determined from the signal. In contrast to a Fourier Transform decomposition where the frequencies of the sinusoidal signals are predetermined in the algorithm and do not change with the signal being processed, the frequencies calculated by the preferred signal processing algorithm are based on the signal.

[0016] Alternatively, the preferred signal processing algorithm represents the signal as one or a sum of alternating signal(s) having an instantaneous amplitude and an instantaneous frequency that vary in time more slowly than a pure sinusoidal signal at the instantaneous frequency. ii) The preferred signal processing algorithm can extract out the natural roll period above other potential competing periods in the signal, typically associated with sea state or heave.

[0017] More particularly, the sea state or heave component in the signal is preferably filtered out before the preferred signal processing algorithm is applied. For example, a high-pass filter can be used to remove the low-frequency components associated with the sea state. The cutoff frequency should be set based on the typical frequencies of the wave-induced motions. The higher frequency components, which are more likely related to the vessel's own dynamics, are retained. The preferred signal processing algorithm can perform a spectral analysis of the IMU data to identify and isolate the cutoff frequency to be used to filter out the sea state or heave component in the signal. Tools like the Fast Fourier Transform (FFT) can be used for this limited purpose.

[0018] Alternatively, wave spectra shapes are predictable: there are empirical formulas for the wave spectra or they can be measured from the weather forecast or wave radar. This signal processing algorithm can compute the Response Amplitude Operators (“RAO”), which indicate thevessel’s dynamics, including the natural roll period, from the wave spectrum calculated via an empirical formula. The computed RAO shows the vessel's characteristics, including the natural roll period.

[0019] Alternatively yet, a spectrogram or waterfall plot can be used to analyze the frequency content of the time signal of the vessel’s roll. These plots can help identify the natural roll period even when certain frequencies (like those of the sea state or heave) dominate the signal. iii) The preferred signal processing algorithm can extract the natural roll period in a manner in which it does not require an excessive number of data points.

[0020] More particularly, signal processing algorithms, such as Prony’s method, or Systems Identification algorithms (e.g., Auto-Regressive Moving Average, or similar) can detect the frequencies in a signal based on very short time signals. The duration of the time series can be very short - generally just like in the range approximately between one fifth of and one full period. As used herein, approximately means + / - 10 percent.

[0021] When the roll angles become large enough, the container stacks will also start to vibrate, which will be detectable in both the lashing loads and the accelerations that can be measured at the top of the stacks. In some embodiment, it is possible to use the equipment and algorithms described in U.S. application serial no. 18 / 049,249, filed on 24 October 2022, to provide supplemental means of detecting larger, potentially more dangerous conditions. U.S. application serial no. 18 / 049,249 describes a monitoring system designed to extract critical natural frequencies of container stack vibrations. For example, the monitoring system described in U.S. application serial no. 18 / 049,249 and the monitoring system described herein above can be used together with a "voting" type algorithm where both systems can agree before an alarm or an alarm level is triggered.

[0022] Optionally, the monitoring system described herein can also be used to detect the vessel's torsional modes and track when there is a significant build-up of torsional vibration amplitude, although it could be more effective to have this function performed by one or more accelerometer(s) that is(are) tied to the vessel deck or hatch covers.

[0023] Furthermore, the disclosure generally describes a monitoring system for a parametric roll risk prediction. The risk predictions are based on the natural roll period that has been detected asexplained herein above. The risk predictions can warn the vessel operators of the potential risk of entering parametric roll, even when the natural roll period of the vessel is relatively constant, sometimes well before the natural roll period shifts during parametric roll of the vessel.

[0024] More particularly, to make / update risk predictions, a risk estimator, in the form of a formula, involving the sea state and the natural roll period of the vessel, is used. The formula may be empirical or derived from a model.

[0025] The monitoring of the natural roll period can be detected using short roll signals as described herein above. The sea state or heave at the location of the vessel can also be measured using short time roll signals at lower frequencies as described herein above. Alternatively, the sea state or heave at the location of the vessel can be estimated, for example, either with an onboard wave radar, or with a third-party weather forecasting service obtained via internet connection and the vessel location (longitude and latitude), heading, and velocity, which can be derived from a global positioning system (“GPS”). The sea state or heave and vessel natural roll period are parameters to be entered into the risk estimator and the subsequent risk plots is calculated. An onsite dashboard is preferably used to display the vessel’s risk of entering parametric roll. The dashboard may show risk levels and uncertainty, at current vessel position.

[0026] For the case where a third-party weather forecasting service is used for the sea state or heave, the vessel’s risk of entering parametric roll at future vessel positions (e.g., 4 to 6 hours) can optionally be estimated and the onsite dashboard can also display the vessel’s future risk based on its current trajectory. A recommendation for vessel path to help reduce the vessel’s risk of entering parametric roll can also be displayed on the onsite dashboard.

[0027] In addition to the parametric roll risk assessment, the monitoring system for parametric roll risk prediction can also estimate the maximum expected roll amplitudes along possible vessel paths when taking into account a model (e.g., RAO).

[0028] Optionally, the monitoring system for parametric roll risk prediction may regularly upload its data (natural roll period, vessel’s position, nearby sea state or heave) to the cloud.

[0029] It is to be understood that the following disclosure describes several exemplary embodiments for implementing different features, structures, or functions of the invention. Exemplary embodiments of components, arrangements, and configurations are described below tosimplify the disclosure; however, these exemplary embodiments are provided merely as examples and are not intended to limit the scope of the invention. Additionally, the disclosure may repeat reference numerals and / or letters in the various exemplary embodiments and across the Figures provided herein. This repetition is for the purpose of simplicity and clarity and does not in itself dictate a relationship between the various exemplary embodiments and / or configurations discussed in the various Figures. Finally, the exemplary embodiments presented below may be combined in any combination of ways, i.e., any element from one exemplary embodiment may be used in any other exemplary embodiment, without departing from the scope of the disclosure.BRIEF DESCRIPTION OF THE DRAWINGS

[0030] For a more detailed description of the embodiments of the disclosure, reference will now be made to the accompanying drawings, wherein:

[0031] FIG. 1 illustrates that vessel’s roll motion spectrum combines the wave spectrum and the Response Amplitude Operators (“RAO”);

[0032] FIG. 2 is a graph that illustrates, with a solid curve, the roll angle in degrees (left axis) of a cargo vessel as a function of time in seconds (horizontal axis) during a parametric roll resonance event, FIG. 2 also illustrates with a dotted the dominant period in seconds (right axis) of the signal within a moving window (e.g., 50 second-duration window or less, such as short as 10 seconds), extracted using Prony's method;

[0033] FIG. 3 shows a schematic of a system that utilizes a risk estimator to predict a vessel’s risk of entering parametric roll using the natural roll period updated in real-time from time series of vessel’s roll;

[0034] FIG. 4 illustrates an example of risk estimator; and

[0035] FIG. 5 shows an example of a dashboard.DETAILED DESCRIPTION

[0036] The natural roll period of a vessel and its cargo refers to the intrinsic roll period of a vessel and its cargo, even in the absence of external forces. It is a theoretical concept that describes how the vessel would roll if it were in water, with no wind, and ideally balanced. This period is determined by the vessel’s design and the cargo it contains, particularly mass distribution and hull geometry, and the location of the waterplane relative to the vessel. It is a fundamental characteristicthat generally changes with the location of the waterplane, and in particular, with the tilt and heigh of the waterplane that is representative of the incident waves.

[0037] This fundamental characteristic affects the Response Amplitude Operators (“RAO”) of the vessel and its cargo. The RAO are functions that describe the motion of the vessel and its cargo in response to the forces that act on them. The RAO can be expressed as a function of time, for example, the RAO can predict the motion of the vessel and its cargo as a function of time in response to the past history of the waves that impinge on it. The RAO can also be expressed as a function of frequency, for example, the RAO can predict the amplitude of the roll angle of the vessel and its cargo as a function of frequency of oscillation in response to the amplitude of the waves that impinge on it at this frequency.

[0038] FIG. 1 illustrates parametric roll in frequency (i.e., with spectra). On the left, FIG. 1 shows a typical spectrum 10 of the force exerted by the waves for a particular sea condition. The spectrum 10 has one peak 12, located at low frequencies, which indicates the dominant rhythm or period of the forces that the waves exert on the vessel and its cargo when the waves impact them.

[0039] On the right, FIG. 1 shows a typical spectrum 14 of the roll angle measured on the vessel. The spectrum 14 illustrates that vessel’s roll motion spectrum combines the wave spectrum and the RAO. Indeed, the spectrum 14 has now two peaks, peak 16, at low frequencies, and peak 18, at higher frequencies. Peak 16 is caused by the dominant rhythm or period of the forces that the waves exert on the vessel and its cargo when the waves impact them. Peak 16 means that the vessel and its cargo will tend to rock at the relatively lower external rhythm or period of the impinging waves. However, peak 18 is caused by the RAO of the vessel and its cargo. Peak 18 means that the vessel and its cargo will also tend to rock at a relatively higher intrinsic rhythm or period.

[0040] As mentioned previously, the intrinsic rhythm or period indicated by peak 18 can change (the rhythm can slow down or period can increase) with the tilt and heigh of the waterplane that is representative of the incident waves. When the intrinsic rhythm become close the external rhythm of the waves, large roll angle may be reached by the vessel and its cargo, causing hazardous conditions for the vessel and its cargo. The lowering of the intrinsic rhythm of the natural vessel’s roll motion toward the external rhythm or period of the impinging waves, and the resulting large roll angles reached by the vessel and its cargo are indicative of parametric roll.

[0041] FIG. 2 illustrates a parametric roll event in time. Note that the natural roll period is generally not the measured roll period (i.e., calculated by zero crossing in a measured time series).

[0042] FIG. 2 shows that before 250 seconds, the dominant period corresponding to the natural roll period of the vessel and its cargo is 30 seconds or less. However, between 250 and 500 seconds, the dominant roll period of the vessel and its cargo shifts (in this case, increases) significantly to 45 seconds or more. Therefore, a shift of the period that is significant can indicate an imminent parametric roll.

[0043] In a preferred embodiment of a vessel monitoring system for rapid detection of parametric roll, an Inertial Measurement Unit (“IMU”) is used to measure the motion of a vessel as a function of time (e.g., at regular time intervals). The IMU is preferably located on the bridge. One IMU is preferred for cost purposes and simplicity, but more IMU can be provided. A person having ordinary skill can expand the disclosure using more than one IMU. The IMU typically provides 3-axis acceleration fused with the 3 angular velocities around those axes.

[0044] The IMU measurement includes the effects of current sea conditions (e.g., waves), wind, vessel loading conditions, and any other external forces acting at the time of measurement. It is a measurement showing the vessel’s behavior under specific conditions. Note that the overall motion of the vessel is not limited to roll, but can include vessel torsion, bending, yaw, etc.

[0045] The roll angle of the vessel and its cargo is then derived, and optionally denoised, from the measurements acquired with the IMU(s). Preferably, a filtering approach, e.g., Madgwick or Kalman or Complementary filter, is used to derive the roll signal from the overall motion of the vessel and its cargo. This roll signal represents the roll motion of the vessel. Statistical methods (e.g., stacking of several measurements performed at nearly the same location on the vessel) can also be used to remove unwanted noise.

[0046] In a preferred embodiment of the vessel monitoring system for rapid detection of parametric roll, the natural roll period is then repeatedly extracted from a subset of the roll signal recently acquired by using a signal processing algorithm.

[0047] The signal processing algorithm starts by filtering the subset of roll signal to filter out the sea state or heave component in the signal. The filtering is a-high pass filter. The cutoff frequency of the filter can be adjusted or tuned based on any prior knowledge of the wave spectrum. In general,band-pass filters can be used to isolate the frequency ranges of interest. This filtering can help in reducing the impact of dominant frequencies outside these ranges, allowing for a clearer analysis of the weaker frequencies.

[0048] The signal processing algorithm then analyzes the filtered subset of the roll signal to identify the dominant roll frequencies (e.g., peak 16 and peak 18 in FIG. 1), and in particular the roll frequency corresponding to the natural roll period (e.g., peak 18 in FIG. 1). An important aspect of the analysis is to detect the natural roll period as early as possible, that is, using the data from a time interval as short as possible. A subset of data collected during an interval spanning from approximatively one fifth of to approximatively one full duration that it took the vessel to effect a full roll cycle, i.e., rolling from one side to the other and back again, can be sufficient. For example, the interval spanning can span approximatively one fourth of a full roll cycle.

[0049] For example, the natural roll period is extracted from the filtered subset via Prony's method or any other system identification algorithm / method. The Prony’s method is preferred since it seems to be the fastest and it can identify the natural roll period in a very short time interval (e.g., less than 50 seconds, as low as 10 seconds). The Prony's method is a parametric modeling technique used to estimate the damping and frequencies in a sum of complex exponential signals. The Prony’s method fits a model consisting of a sum of complex exponential functions to the measured data. The exponent in the complex exponential functions is not predetermined but calculated by the method. Preferably, a smooth window function is applied the signal before performing the Prony’s method. The smooth window function can help reduce the spectral leakage that often causes frequency components to appear stronger or weaker than they actually are.

[0050] However, in general, the natural roll period could be extracted from the roll signal time series via other algorithms. The simplest algorithm is to do a conversion of the time series signal into a spectrum (i.e., Frequency-Domain Analysis (“FDA”)) and then look at what frequency there is a peak. Several other methods can be used to extract the natural roll period, each with its strengths and limitations, including Fourier Analysis, Fast Fourier Transform (“FFT”), Wavelet Transform, Hilbert Transform, Eigenvalue Realization Algorithm (“ERA”), Singular Value Decomposition (“SVD”), various Time-Frequency Analysis methods (e.g., Short Time Fourier Transform (“STFT”) and Continuous Wavelet Transform (“CWT”), with or without normalization).

[0051] The Prony's method can give some residual values indicative of how good the fit compares to the observation. Optionally, the residual values can be used to quantify the natural roll period uncertainty and be translated as percentage of uncertainty. Alternatively, or additionally, a Frequency-Domain Analysis or a Time-Frequency Analysis can be performed to identify dominant frequencies in the IMU data. The inverse of the dominant roll frequency will give another measure of the natural roll period. This other measure of the natural roll period can be compared to the results obtained via Prony’s method to evaluate the natural roll period uncertainty. Alternatively, or additionally, the uncertainty of the natural roll period can be set to cover the range between the calculated value and the calculated value plus an extra few seconds (e.g., approximately 5 seconds). Alternatively, or additionally, the uncertainty of the natural roll period can be set to cover the range of all the values previously calculated during the journey of the vessel.

[0052] The sequence of filtering and applying Prony's method is then repeated on the next subset of roll signal recently acquired.

[0053] In a preferred embodiment of the vessel monitoring system for rapid detection of parametric roll, the natural roll period is tracked and changes of the natural roll period indicate, in general, a change in the vessel dynamics (stiffness, mass, instability due to lack of damping). More particularly, an increase of the natural roll period indicates a risk of parametric roll. Thus, the latest natural roll period is continuously recorded and compared to previous results as it is updated.

[0054] In a preferred embodiment of the vessel monitoring system for rapid detection of parametric roll, the changes of the natural roll period can be automatically detected using a linear fit through the latest the natural roll periods that have been extract. A slope of the linear fit sufficiently large (and positive) can trigger an alarm.

[0055] However, various other ways may be used to automatically detect the changes of the natural roll period, such as Data Segmentation and Stratification, Cross-Condition Analysis, Normalization and Standardization, Statistical Analysis (e.g., Analysis of Variance), Machine Learning and Predictive Modeling.

[0056] In a preferred embodiment of the vessel monitoring system for rapid detection of parametric roll, a monitoring system designed to extract critical natural frequencies of container stack vibrations is also used to trigger an alarm . For example, the monitoring system described in U.S. application serial no. 18 / 049,249 and the monitoring system described herein above can be usedtogether with a "voting" type algorithm where both systems can agree before an alarm or a higher alarm level is triggered.

[0057] However, any Data Fusion and Integration method can be used when more sensors measuring different conditions indicative of excessive roll are available. A Data Fusion and Integration method combines data from different sensors to create a more comprehensive dataset. This dataset can provide a more holistic view and can be particularly effective if the different conditions cover complementary aspects of excessive roll.

[0058] In a preferred embodiment of the vessel monitoring system, the functionality of the system is extended to the prediction of parametric roll. To do so, a risk estimator, in the form of a formula, involving the sea state or heave and the natural roll period of the vessel, is used. The formula may be empirical or derived from a model. An example of formula can be found in Roll Risk Estimator VI.3 , downloadable from https: / / www.marin.n1 / en / jips / toptier#notice.

[0059] FIG. 3 shows a schematic of a system that utilizes a risk estimator to predict a vessel’s risk of entering parametric roll using the natural roll period updated in real-time from time series of vessel’s roll. An IMU 20 is preferably located on the bridge, measures motion data, and provides the data to a computer 22. The computer 22, for example, a personal computer, is used for processing the data measured by the IMU 20 and extract the natural roll period of the vessel and its uncertainty using a signal processing algorithm describes herein above. Furthermore, the computer 22 can interrogate a weather service 24 residing in the cloud to estimate the sea state or heave (e.g., mean wave period or wave encounter period, effective wavelength, wave direction), or a radar 26 can provide measurements that are then converted into an estimated sea state or heave by the computer 22.

[0060] An example of risk estimator is illustrated in FIG. 4. In a preferred embodiment of the vessel monitoring system, the input of the risk estimator are automatically entered in real-time as follows. The risk estimator then return a risk level on parametric roll, either corresponding to current data, or future data.The current or future sea state or heave is estimated by the computer 22 from the weather service 24 or data from the radar 26.In contrast with the prior art, where the natural roll period and its uncertainty are usually determined using some empirical formulas from vessel building, the natural roll period and its uncertainty calculated as described herein above are also automatically entered in real-time. It has been determined that the natural roll period changes with loading and draft condition of the vessel and that it is preferable to update the natural roll period frequently based on the latest available measurements.The current vessel speed and vessel course are derived from GPS data. The future vessel speed and vessel course are entered manually.

[0061] In a preferred embodiment of the vessel monitoring system for the prediction of parametric roll, Interactive Visualization tools are used. The Interactive Visualization tool is used to explore data interactively. This can help in identifying patterns or anomalies that might not be evident through automated analysis alone. An example of Interactive Visualization tool is shown in FIG. 5. The Interactive Visualization tool can display the current risk level of parametric roll, based on nearby sea state or heave, current vessel speed, and current vessel course. The Interactive Visualization tool can also display the fictive risk level of parametric roll, based on future sea state or heave, and / or future vessel speed, and future vessel course.

[0062] In a preferred embodiment of the vessel monitoring system for the prediction of parametric roll, Robustness and Sensitivity Analysis is provided. The Analysis shows how changes in conditions affect your models or conclusions. This Analysis helps in understanding the robustness of the risk level computed by the risk estimator and in identifying conditions that have a significant impact. For example, the natural roll period can be altered automatically by an extra few seconds (e.g., approximately 5 seconds, or to a value previously calculated during the journey of the vessel) and the impact of the risk level recomputed. When the impact of the risk level is large, the selection of the vessel speed and course may be analyzed more closely.

[0063] In addition to the foregoing, the disclosure also contemplates at least the following embodiments 1-16. It should be noted that any element of these embodiments may further include details related to this element that are disclosed in a paragraph or Figure describing the preferred embodiments without necessarily including details of other elements that are disclosed in the same or other paragraph or Figure.Embodiment 1

[0064] Embodiment l is a monitoring system for the rapid detection of parametric roll of vessels. The monitoring system comprises an Inertial Measurement Unit (“IMU”) fixed to the vessel in a convenient location (e.g., bridge, wheelhouse, etc.). The monitoring system comprises a computer for processing data. The IMU measures vessel motion and transmits vessel motion data (i.e., raw data) to the computer.

[0065] The computer is programmed to compute a roll signal (i.e., processed data) from the vessel motion data, for example, e.g., by filtering the raw data using either Kalman or Complementary Filtering to ensure that the signal being processed in subsequent steps is as clean as possible. Otherwise, the raw data can be used without this computational step.

[0066] Furthermore, the computer is generally programmed to run an algorithm that can repeatedly detect the periods or frequencies of multiple individual sine waves composing the roll signal during short time intervals. More specifically, the duration of the short time intervals is preferably between approximately one fifth and approximately one full of the duration of a roll cycle of the vessel. The duration of the short time intervals is more preferably approximately one fourth of the duration of a roll cycle of the vessel.

[0067] Further, the computer is generally programmed to repeatedly identify one of the periods or frequencies of the multiple individual sine waves as the natural roll period or frequency of the vessel. More specifically, the periods or frequencies of the multiple individual sine waves corresponding to relatively longer periods or relatively lower frequencies are preferably rejected as the roll period or frequency corresponding to sea state or heave, and the one of the periods or frequencies of the multiple individual sine waves corresponding to relatively shorter periods or relatively higher frequencies is preferably identified as the natural roll period or frequency of the vessel. The one of the periods or frequencies of the multiple individual sine waves in a predetermined range is more preferably identified as the natural roll period or frequency of the vessel.

[0068] Further, the computer is generally programmed to trigger an alarm upon a significant shift in the vessel roll natural period or frequency. More specifically, the amount that qualifies for a significant shift can be determined using simulations and / or experimental data on any vessel and its cargo. Additionally, or alternatively, a shift that is considered significant can correspond to aroll having a sufficiently large amplitude (e g., occurs or has occurred contemporarily with roll angles larger than approximately ten degrees).Embodiment 2

[0069] Embodiment 2 is a monitoring system as described in embodiment 1 wherein the algorithm generally includes a parametric modeling technique used to estimate the damping and frequencies in a sum of complex exponential signals, and more specifically, implements the Prony’s method.Embodiment 3

[0070] Embodiment 3 is a monitoring system as described in embodiment 1 wherein the algorithm generally includes a computation of a signal in quadrature with the roll signal, and more specifically a Hilbert transform of the roll signal. The algorithm also includes a computation of an instantaneous frequency and / or amplitude using the signal in quadrature with the roll signal. The instantaneous amplitude and an instantaneous frequency typically vary in time more slowly than a pure sinusoidal signal at the instantaneous frequency.Embodiment 4

[0071] Embodiment 4 is a monitoring system as described in embodiment 1 wherein the algorithm generally includes a computation of an Auto Regressive Moving Average (“ARMA”) of the roll signal.Embodiment 5

[0072] Embodiment 5 is a monitoring system as described in any of embodiments 1 to 4 wherein the alarm is generally triggered upon the significant shift in the vessel roll natural period or frequency and a value of a heave period of frequency nearby the vessel. More specifically, the alarm is triggered upon the vessel roll natural period or frequency being less than 10% higher than the value of the heave period of frequency nearby the vessel.Embodiment 6

[0073] Embodiment 6 is a monitoring system as described in any of embodiments 1 to 5 further comprising a monitoring system generally designed to extract critical natural frequencies of container stack vibrations and the alarm is generally triggered upon the significant shift in the vesselroll natural period or frequency and the values of the extracted critical natural frequencies of the container stack vibrations. More specifically, the alarm is triggered using a "voting" type algorithm involving the monitoring system for the rapid detection of parametric roll of vessels and the monitoring system designed to extract critical natural frequencies of the container stack vibrations.Embodiment 7

[0074] Embodiment 7 is a method of using a monitoring system as described in any of embodiments 1 to 6. The method comprises the steps of: measuring vessel motion with the IMU (i.e., raw data); transmitting vessel motion data to the computer; computing a roll signal (i.e., processed data) from the vessel motion data (e.g., by filtering the raw data using either Kalman or Complementary Filtering to ensure that the signal being processed in subsequent steps is as clean as possible), otherwise the raw data can be used without this computational step; repeatedly detecting the periods or frequencies of multiple individual sine waves composing the roll signal during short time intervals; repeatedly identifying one of the periods or frequencies of the multiple individual sine waves as the natural roll period or frequency of the vessel; and triggering an alarm upon a significant shift in the vessel roll natural period or frequency.Embodiment 8

[0075] Embodiment 8 is a monitoring system for estimating a vessel’s risk of entering parametric roll.

[0076] The monitoring system comprises a device onboard the vessel that is generally capable of accurately measuring the vessel’s roll at time intervals, and more specifically, an IMU.

[0077] The monitoring system comprises a computer for running a signal processing algorithm configured to estimate the natural roll period or frequency of the vessel in real-time from the measured the vessel’s roll.

[0078] The monitoring system either comprises a sensor adapted to measure the sea state or heave nearby the current vessel position and / or nearby the future vessel position or comprises a data connection for downloading the sea state or heave data.

[0079] The monitoring system comprises a risk estimator generally allowing to compute the vessel’s risk of entering parametric roll as a function of the sea state or heave data, the natural roll period or frequency of the vessel, and the vessel trajectory.Embodiment 9

[0080] Embodiment 9 is a monitoring system as described in embodiment 9, wherein the signal processing method involves the Prony’s method.Embodiment 10

[0081] Embodiment 10 is a monitoring system as described in embodiments 8 or 9, wherein the sea state or heave is determined by either an onboard wave radar or downloaded from a third-party weather forecasting service.Embodiment 11

[0082] Embodiment 11 is a monitoring system as described in any of embodiments 8 to 10, wherein the regional sea state is used to forecast the vessel’s risk of entering parametric roll in the area surrounding the current position of the vessel.Embodiment 12

[0083] Embodiment 12 is a monitoring system as described in any of embodiments 8 to 11, further comprising a dashboard displaying an alarm when the vessel is deemed to be at high risk of entering parametric roll.Embodiment 13

[0084] Embodiment 13 is a monitoring system as described in any of embodiments 8 to 12, in combination with is a monitoring system as described in any of embodiments! to 6.Embodiment 14

[0085] Embodiment 14 is a method of using a monitoring system as described in any of embodiments 8 to 13.

[0086] The method comprises the steps of measuring the vessel’s roll at time intervals; estimating the natural roll period of the vessel in real-time from the measured the vessel’s roll; either measuring sea state or heave data or downloading sea state or heave data; and computing the vessel’s risk of entering parametric roll as a function of the sea state or heave data, the natural roll period of the vessel, and the vessel trajectory.Embodiment 15

[0087] Embodiment 15 is a method as described in embodiment 14 further comprising comparing the vessel’s risk of entering parametric roll for at least two vessel paths and optionally making a recommendation for a path of the vessel.

Claims

What is claimed is:

1. A monitoring system for the rapid detection of parametric roll of vessels, comprising: a device onboard the vessel that is capable of measuring a roll signal indicative of the vessel motion at time intervals; a computer for processing data; wherein the device is configured to transmit roll signal data to the computer; wherein the computer is programmed to: repeatedly detect the periods or frequencies of multiple individual sine waves composing the roll signal during short time intervals, repeatedly identify one of the periods or frequencies of the multiple individual sine waves as the natural roll period or frequency of the vessel, and trigger an alarm upon a significant shift in the vessel roll natural period or frequency.

2. The monitoring system of claim 1, wherein the program to detect the periods or frequencies of multiple individual sine waves composing the roll signal during short time intervals includes: a parametric modeling technique used to estimate the damping and frequencies in a sum of complex exponential signals, or a computation of a signal in quadrature with the roll signal, and a computation of an instantaneous frequency and amplitude using the signal in quadrature with the roll signal, or a computation of an Auto Regressive Moving Average (“ARMA”) of the roll signal.

3. The monitoring system of claim 1, wherein the program to trigger the alarm further compares a value of a heave period of frequency nearby the vessel to a threshold to trigger the alarm.

4. The monitoring system of claim 1, further comprising a monitoring system generally designed to extract critical natural frequencies of container stack vibrations and wherein the program to trigger the alarm compares a value of one of the extracted critical natural frequencies of the container stack vibrations to a threshold.

5. The monitoring system of claim 1, wherein the device includes an IMU fixed to the vessel.

6. A method of using the monitoring system of any of claims 1 to 5, comprising: measuring the roll signal with the device; transmitting roll signal data to the computer; repeatedly detecting the periods or frequencies of multiple individual sine waves composing the roll signal during short time intervals; repeatedly identifying one of the periods or frequencies of the multiple individual sine waves as the natural roll period or frequency of the vessel; and triggering an alarm upon a significant shift in the vessel roll natural period or frequency.

7. The method of claim 6, further triggering the alarm upon the vessel roll natural period or frequency being less than 10% higher than the value of the heave period of frequency nearby the vessel.

8. The method of claim 6, wherein the alarm is triggered using a voting type algorithm involving the monitoring system for the rapid detection of parametric roll of vessels and the monitoring system designed to extract critical natural frequencies of the container stack vibrations.

9. A monitoring system for estimating a vessel risk of entering parametric roll, comprising: a device onboard the vessel that is capable of measuring a roll signal indicative of the vessel motion at time intervals; a computer for processing data; wherein the device is configured to transmit roll signal data to the computer;wherein the computer is programmed to: repeatedly detect the periods or frequencies of multiple individual sine waves composing the roll signal during short time intervals, and repeatedly identify one of the periods or frequencies of the multiple individual sine waves as the natural roll period or frequency of the vessel; a sensor adapted to measure the sea state or heave nearby the current vessel position and / or nearby the future vessel position, or a data connection for downloading the sea state or heave data; and a risk estimator configured to compute the vessel risk of entering parametric roll as a function of the sea state or heave data, the natural roll period or frequency of the vessel, and a vessel trajectory.

10. The monitoring system of claim 9, wherein the program to detect the periods or frequencies of multiple individual sine waves composing the roll signal during short time intervals includes: a parametric modeling technique used to estimate the damping and frequencies in a sum of complex exponential signals, or a computation of a signal in quadrature with the roll signal, and a computation of an instantaneous frequency and amplitude using the signal in quadrature with the roll signal, or a computation of an Auto Regressive Moving Average (“ARMA”) of the roll signal.

11. The monitoring system of claim 10, wherein the sea state or heave is determined by either an onboard wave radar or downloaded from a third-party weather forecasting service.

12. The monitoring system of claim 12, further comprising a dashboard displaying an alarm when the vessel is deemed to be at high risk of entering parametric roll.

13. A method of using the monitoring system of any of claims 9 to 12, comprising measuring the vessel’s roll at time intervals;estimating the natural roll period of the vessel in real-time from the measured the vessel’s roll; either measuring sea state or heave data or downloading sea state or heave data; and computing the vessel risk of entering parametric roll as a function of the sea state or heave data, the natural roll period of the vessel, and a vessel trajectory.

14. The method of claim 13 further comprising using regional sea state or heave data to forecast the vessel risk of entering parametric roll in an area surrounding the current position of the vessel.

15. The method of claim 13 further comprising comparing the vessel’s risk of entering parametric roll for at least two vessel paths to make a recommendation for a path of the vessel.

Images