Method for monitoring long-distance communication fault of underwater cleaning robot

Abstract

This invention relates to the field of underwater equipment communication transmission technology, specifically to a method for long-distance communication fault monitoring of underwater cleaning robots. The invention first controls the robot to execute preset actions and acquire a response vector under a communication early warning state; then, it performs time delay alignment based on a reference response vector and calculates the channel change vector by comparing the aligned response vectors; further, it obtains the electromagnetic interference energy ratio based on the energy distribution characteristics of the channel change vector before the peak arrives; further, it obtains the channel response sensitivity based on the energy difference characteristics between the channel change vector and the reference response vector; finally, it identifies communication faults and issues control commands based on the electromagnetic interference energy ratio and channel response sensitivity. This not only improves system robustness but also ensures the continuity of deep-sea operations and equipment safety.

Description

Technical Field

[0001] This invention relates to the field of underwater equipment communication transmission technology, specifically to a method for long-distance communication fault monitoring for underwater cleaning robots. Background Technology

[0002] In deep-sea operations such as offshore platform cleaning and ship maintenance, underwater cleaning robots are typically powered by umbilical cables and rely on underwater acoustic communication to interact with surface control consoles. However, during high-voltage frequency conversion-driven cleaning pump operations, the communication link often experiences increased bit error rates or even intermittent interruptions due to concurrent interference from various physical factors. These interference sources include weak radio frequency leakage caused by aging umbilical cable insulation, scattering and attenuation of sound waves by cavitation bubbles, the robot entering multipath interference dead zones, and attitude drift caused by recoil forces. Because these faults exhibit similar characteristics in traditional indicators such as received signal strength or signal-to-noise ratio, existing technologies struggle to accurately identify their specific causes in the early stages of communication degradation. Summary of the Invention

[0003] To address the technical problem of being unable to accurately distinguish different fault sources based solely on physical layer signal characteristics during periods of deteriorating communication quality, the present invention aims to provide a method for long-distance communication fault monitoring specifically for underwater cleaning robots. The technical solution adopted is as follows:

[0004] In the communication warning state, the load driving current of the robot is controlled to perform preset steady-state holding, lower step perturbation and recovery actions in sequence, and the underwater acoustic channel impulse response at each stage is obtained, which are the reference response vector, lower step response vector and recovery response vector, respectively.

[0005] Using the reference response vector as a reference, the lower step response vector and the recovery response vector are time-delay aligned respectively; the channel change vector is calculated by comparing the aligned response vectors; the electromagnetic interference energy ratio is obtained based on the energy distribution characteristics of the channel change vector before the main peak arrives; and the channel response sensitivity is obtained based on the energy difference characteristics between the channel change vector and the reference response vector.

[0006] Based on the electromagnetic interference energy ratio and the channel response sensitivity, communication faults are identified and control commands are issued.

[0007] Furthermore, the method for obtaining the channel change vector includes:

[0008] Based on the phase correlation between the reference response vector and the lower step response vector, either the complex difference mode or the amplitude difference mode is selected.

[0009] Based on the selected differential mode, the channel change vector is calculated by comparing the aligned response vector.

[0010] Furthermore, the method for selecting the difference mode includes:

[0011] The magnitude of the normalized complex dot product of the aligned lower step response vector and the reference response vector is calculated and used as the phase correlation coefficient. The phase correlation coefficient is compared with a preset correlation threshold, and either the complex difference mode or the amplitude difference mode is selected.

[0012] Furthermore, when the complex differential mode is selected, the aligned lower step response vector and the reference response vector are differentially analyzed element by element to obtain the channel change vector;

[0013] When the amplitude differential mode is selected, the magnitude of each component in the aligned lower step response vector is differentially divided element-wise with the magnitude of each component in the reference response vector to obtain the channel change vector.

[0014] Furthermore, the method for obtaining the electromagnetic interference energy ratio includes:

[0015] The point with the largest modulus of the component in the reference response vector is extracted as the main peak point, and the point with the preset protection interval before the main peak point is taken as the early arrival cutoff point to obtain the early arrival interval.

[0016] The electromagnetic interference energy ratio is obtained by comparing the energy of the channel change vector within the early arrival interval with the total energy of the channel change vector.

[0017] Furthermore, the method for determining communication faults includes:

[0018] When the electromagnetic interference energy ratio is greater than a preset safe energy threshold, it is determined to be an electromagnetic leakage fault;

[0019] When the electromagnetic interference energy ratio is less than or equal to a preset safe energy threshold, and the channel response sensitivity is less than a preset response correlation threshold, it is determined to be a multipath dead zone fault.

[0020] Furthermore, the control commands include:

[0021] In response to the electromagnetic leakage fault, the controller immediately sends a command to the power management module to disconnect the relay or contactor of the main power supply circuit of the cleaning pump motor;

[0022] In response to the multipath dead zone fault, the controller sends a command to the walking driver to drive the robot to perform an escape and relocation action.

[0023] Furthermore, when the complex difference mode is selected, the method further includes: performing element-wise difference between the aligned lower step response vector and the recovery response vector to obtain the recovery change vector.

[0024] Furthermore, based on the real part of the normalized complex dot product of the channel change vector and the recovery change vector, the change closure index is obtained, and the communication fault is further determined according to the change closure index.

[0025] Furthermore, when the change closure index is greater than a preset fluid reversibility threshold, it is determined to be a bubble blockage fault; when the change closure index is less than or equal to the preset fluid reversibility threshold, it is determined to be a mechanical drift fault.

[0026] The present invention has the following beneficial effects:

[0027] This invention first controls a robot to execute preset actions and acquire response vectors under communication early warning conditions, providing a basis for subsequent analysis. To further avoid phase mismatch caused by sampling time errors, which could introduce false dynamics into subsequent differential results and mask real physical changes, the lower step response vector and the recovery response vector are time-delay aligned using a reference response vector. The aligned response vectors are then compared to calculate the channel change vector, extracting the dynamic increment caused solely by current excitation, thereby improving sensitivity to weak or hidden faults. Furthermore, based on the energy distribution characteristics of the channel change vector before the main peak arrives, the electromagnetic interference energy ratio is obtained as a basis for subsequent determination of the degree of electromagnetic leakage. Additionally, based on the energy difference characteristics between the channel change vector and the reference response vector, the channel response sensitivity is obtained as a supplementary basis for subsequent fault determination. Finally, based on the electromagnetic interference energy ratio and the channel response sensitivity, communication faults are identified and control commands are issued. In the communication early warning state, this invention excites the channel response by current perturbation, decouples the fault source based on the energy distribution before the main peak and the response sensitivity, and realizes accurate identification and self-healing control of faults such as electromagnetic leakage and multipath dead zone, thus solving the technical problem of being unable to distinguish the causes of deterioration. Attached Figure Description

[0028] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0029] Figure 1 A flowchart illustrating a method for long-distance communication fault monitoring of an underwater cleaning robot, as provided in an embodiment of the present invention;

[0030] Figure 2 A flowchart illustrating a method for identifying communication faults, provided as an embodiment of the present invention. Detailed Implementation

[0031] To further illustrate the technical means and effects adopted by the present invention to achieve its intended purpose, the following, in conjunction with the accompanying drawings and preferred embodiments, details the specific implementation, structure, features, and effects of a method for long-distance communication fault monitoring of underwater cleaning robots proposed according to the present invention. In the following description, different "one embodiment" or "another embodiment" do not necessarily refer to the same embodiment. Furthermore, specific features, structures, or characteristics in one or more embodiments can be combined in any suitable form.

[0032] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this invention pertains.

[0033] The following description, in conjunction with the accompanying drawings, details a specific solution for a long-distance communication fault monitoring method for underwater cleaning robots provided by this invention.

[0034] Please see Figure 1 The diagram illustrates a flowchart of a method for long-distance communication fault monitoring for underwater cleaning robots, provided by an embodiment of the present invention, specifically including:

[0035] Step S1: In the communication warning state, control the robot's load drive current to sequentially perform preset steady-state hold, lower step perturbation and recovery actions, and obtain the underwater acoustic channel impulse response at each stage, which are the reference response vector, lower step response vector and recovery response vector, respectively.

[0036] This invention is applicable to monitoring communication faults in underwater robots (such as underwater cleaning) powered by umbilical cables and equipped with underwater acoustic communication devices. It is activated when the robot is in cleaning load operation (i.e., the cleaning pump drive current is not less than 80% of the rated value and it is in cavitation jet operation) and the communication link deteriorates but is not completely interrupted.

[0037] In one embodiment of the present invention, the triggering conditions for the communication warning state include, but are not limited to, any of the following situations lasting for more than a preset time window (e.g., 500ms): (1) the bit error rate (BER) output by the physical layer of the underwater acoustic communication device exceeds (2) The link layer automatic repeat request (ARQ) frequency exceeds 3 times per second; (3) The received signal-to-noise ratio (SNR) is less than 10dB; (4) The application layer heartbeat packet loss rate exceeds 50%.

[0038] These metrics can be derived from the communication device registers, protocol stack statistics, or higher-level estimations. Once any trigger condition is met, the system initiates the admission judgment of the fault monitoring process: if at least one channel impulse response (CIR) can be successfully obtained through the data frame synchronization signal (such as the preamble capture pulse) within a subsequent preset window (e.g., 50ms), it is determined that the system has entered the communication warning state;

[0039] If the synchronization signal is missing, the local timer forced acquisition mechanism is automatically activated. The controller directly sends a "forced acquisition" command to the receiver's analog-to-digital converter (ADC) module or baseband processing unit to capture a frame of time-domain signal at the current moment and perform related calculations, or directly reads the latest noise floor data from the receiver's buffer. As long as physical layer data (including pure noise spectrum) that can be used for energy analysis can be obtained, it is considered that the monitoring conditions are met, the system suspends the normal operation logic, and switches to the fault monitoring process of this embodiment of the invention.

[0040] Although this method mainly acquires the ambient noise spectrum when there is no signal, the noise spectrum still contains possible in-band electromagnetic interference characteristics (manifested as an overall increase in the noise floor), so it is still effective for subsequent diagnosis of electromagnetic leakage faults; however, pure noise does not have an effective main peak, so the system marks the current sampling state as "asynchronous noise mode", and skips some fine calculations based on the flag state (explained in the following steps).

[0041] In the communication warning state, in order to decouple fluid factors (cavitation bubbles) and rigid body factors (mechanical drift) in a single underwater acoustic channel, the controller is configured to send commands to the variable frequency drive of the cleaning pump to control the robot's load drive current to sequentially perform preset steady-state hold, lower step perturbation and recovery actions, and obtain the underwater acoustic channel impulse response at each stage to provide a basis for subsequent analysis.

[0042] In a preferred embodiment of the present invention, the fault monitoring time is set to zero, and the first stage is the steady-state maintenance stage (from 0 to 1). During this phase, the controller instructs the cleaning pump to maintain its current set operating current (e.g., 100% of the rated current) to bring the underwater robot into a normal operational equilibrium state. This operational equilibrium state serves as a baseline reference point for fault diagnosis, used to capture the raw channel characteristics at the time of the fault. The duration of this phase must be sufficient for the channel estimator to complete at least one stable background noise or signal sampling, for example, no less than 200 milliseconds.

[0043] The second stage is the next step perturbation stage. arrive , The controller commands the drive current to decrease stepwise to a preset perturbation value (e.g., 80%~90% of the rated current), and starts a timer to maintain this state for a specific perturbation window duration. In this embodiment, Set to 300 milliseconds. The selection of this timeframe is based on specific physical constraints: on the one hand, 300 milliseconds is significantly longer than the dissipation and stabilization time of the cavitation bubble cloud (usually about 50 milliseconds), which is sufficient to cause changes in bubble density and sound wave scattering characteristics; on the other hand, this timeframe is much shorter than the mechanical inertial response time of the underwater robot (usually greater than 1 second), during which the displacement or attitude change of the robot due to the reduction in recoil force is minimal, and it can be approximated that the physical position remains unchanged. Therefore, this stage mainly introduces changes in the fluid field, while shielding changes in the rigid body position.

[0044] The third stage is the current recovery (recovery) stage. arrive , When the timer reaches... Afterwards, the controller commands the drive current to step back to the initial steady-state setpoint and maintain this position for a fluid rebuilding period. (e.g., 100 milliseconds). The current recovery phase is used to verify the reversibility of the channel state: if the change is caused by a bubble, the bubble cloud will be quickly rebuilt as the current recovers, and the channel state should return to the baseline; if the change is caused by mechanical drift, the position cannot be automatically reset in a short time due to inertia and fluid resistance, and the channel state will exhibit an irreversible shift.

[0045] The system detects signals at various stages: from 0 to... Inside, the controller reads the impulse response of the underwater acoustic channel and stores it as a reference response vector. The response vector is a complex array that records the amplitude and phase information of the sound wave along multiple paths.

[0046] exist arrive Inside, the controller performs trigger detection and reading operations again, storing the data as the next step response vector. This vector reflects the channel state when the bubble density decreases but the mechanical position remains essentially unchanged.

[0047] exist arrive Inside, the controller performs trigger detection and reading operations again, storing the data as a recovery response vector. This is used for subsequent analysis of the reversibility of physical processes.

[0048] It should be noted that the thresholds and parameters set during the data acquisition phase can be adaptively adjusted or recalibrated by the implementer according to the needs of the scenario. For example, for heavy-duty robots with high inertia, The duration of the perturbation window can be extended. The setting should be greater than the single-frame data transmission duration of the underwater acoustic communication device or the channel estimation update cycle; alternatively, the controller needs to parse the timing of the communication protocol and synchronize the start time of the next step perturbation action with the pilot time of the communication data frame to ensure that... At least one valid channel impulse response can be captured within the window.

[0049] It should be noted that in other embodiments of the present invention, in order to prevent random errors in a single sampling, the robot can be controlled to repeatedly perform the data acquisition process (e.g., 3 times), and the vector data acquired each time can be cached. Subsequently, a module can be set to perform averaging or optimization processing (e.g., selecting the set of data with the highest signal-to-noise ratio).

[0050] Step S2: Using the reference response vector as a reference, align the lower step response vector and the recovery response vector with time delay; compare the aligned response vectors to calculate the channel change vector; obtain the electromagnetic interference energy ratio based on the energy distribution characteristics of the channel change vector before the main peak arrives; obtain the channel response sensitivity based on the energy difference characteristics between the channel change vector and the reference response vector.

[0051] Before performing complex signal processing, the acquired raw vector data is first screened for validity to remove invalid data caused by equipment malfunctions or extreme dead zones.

[0052] In a preferred embodiment of the present invention, the average energy value of each vector, that is, the average value of the magnitude of each component, is used as the average energy value. The noise floor energy threshold is determined based on the statistical noise floor level of the communication device in silent state, for example, twice the average noise floor value.

[0053] If the average energy value of any vector being analyzed is lower than the noise floor energy threshold, it indicates that the receiver has failed to capture any valid acoustic or electromagnetic signals at the current moment. The system directly marks the sampling as invalid and interrupts the current diagnostic process, or directly jumps to the "multipath dead zone" determination logic (because extremely low energy is itself one of the characteristics of the deep-sea multipath dead zone).

[0054] Meanwhile, if the sampling state is marked as "asynchronous noise mode" (i.e., no effective pilot signal was captured, only the noise floor or time domain signal was acquired), the system skips the time delay alignment, main peak localization, and feature extraction steps based on fine time domain structure (including subsequent calculation of electromagnetic interference energy ratio and change closure exponent), and instead performs a coarse screening process based on total energy anomaly detection:

[0055] The total energy of the current noise floor is calculated and compared with the statistical average of the noise floor energy during normal operation. If the total energy increases significantly (e.g., exceeding the average plus 3 times the standard deviation), an electromagnetic leakage is initially determined, triggering a first-level protection command. Otherwise, considering the continuous deterioration of communication errors, a multipath dead zone is identified, and a second-level escape action is executed. Although it cannot distinguish between air bubbles and mechanical drift, it can ensure basic identification capability for the most dangerous fault (electromagnetic leakage) under complete loss of synchronization conditions.

[0056] In deep-sea operations, underwater acoustic channels are affected by receiver clock drift, Doppler frequency shift, and platform micro-shaking, resulting in non-physical offsets in the channel impulse response acquired at different times. Directly comparing the original vectors can lead to phase mismatch due to sampling time misalignment, causing spurious dynamics to be mixed into the subsequent differential results and masking the real physical changes. Therefore, using the reference response vector as a benchmark, the next step response vector and the recovery response vector are time-delay aligned separately.

[0057] Active step adjustment of the cleaning pump current can disturb the fluid field or mechanical state, thereby changing the underwater acoustic propagation environment. However, this change is implicit in the absolute channel response and is difficult to identify. Therefore, by comparing the aligned response vectors to calculate the channel change vector, the dynamic increment caused only by current excitation can be extracted, thereby improving the sensitivity to weak or hidden faults.

[0058] Electromagnetic waves propagate along the umbilical cable at near the speed of light, while sound waves travel at only about 1500 m / s in water, a difference of orders of magnitude. Radio frequency leakage generated by frequency conversion drives typically manifests as broadband impulse noise or non-stationary interference. Although this type of interference cannot form a coherent main peak in the channel impulse response similar to that of sound waves, because it propagates at near the speed of light through the umbilical cable, it arrives at the receiver front-end earlier than the underwater acoustic signal. This causes a significant increase in the noise floor energy within the "acoustic silence zone" before the main peak of the channel impulse response, differing from pure Gaussian white noise background.

[0059] Therefore, based on the energy distribution characteristics of the channel change vector before the main peak arrives, the electromagnetic interference energy ratio is obtained as the basis for subsequent judgment of the degree of electromagnetic leakage. Furthermore, since it is directly related to hardware hazards such as insulation aging and is not affected by acoustic factors such as bubbles and position, it solves the blind zone problem that traditional leakage protection cannot detect high-frequency weak radio frequency leakage.

[0060] If the communication degradation originates from the robot's spatial location (such as a multipath dead zone), then changes in the cleaning pump's current will not significantly alter the channel characteristics—because the sound field is determined by the geometric path and is independent of the fluid. Conversely, if the degradation is caused by the cleaning operation itself (such as bubble scattering or recoil force), the channel will respond significantly to current disturbances. Therefore, based on the energy difference characteristics between the channel change vector and the baseline response vector, the channel response sensitivity is obtained as a supplementary basis for subsequent fault determination.

[0061] Preferably, in one embodiment of the present invention, a cross-correlation algorithm is used to align the main peaks of the lower step response vector and the recovery response vector with the reference response vector, respectively.

[0062] As an example, let's set the sampling rate to 100kHz and the sliding delay variable... The range of values ​​is In this example, W is set to 10, corresponding to 10 sampling points; Calculate the cross-correlation function of the two vectors within the range, find the time delay value corresponding to the maximum cross-correlation magnitude, and this time delay value is the time deviation to be corrected.

[0063] It should be noted that the specific value of W can be adjusted by the implementer according to the clock stability index of the receiver hardware. The specific calculation process of cross-correlation peak search and delay estimation and cross-correlation alignment belongs to the existing technology in the field of digital signal processing and will not be elaborated here.

[0064] However, unlike the cyclic prefix (CP) processing in conventional communication equalization, this invention strictly adopts a "linear shift and zero-padding" strategy when shifting and correcting the vector based on the calculated delay value (when it is determined from the cross-correlation results that the vector needs to be shifted forward or backward). When sampling a number of points, zero values ​​are directly filled at the beginning or end of the time-domain sequence to achieve shifting (without connecting the beginning and end of the sequence). Circular shifting is strictly prohibited to ensure strict preservation of time-domain causality and prevent non-physical winding of signal energy on the time axis.

[0065] After timing alignment, before calculating the channel change vector, considering that even if the phase is aligned, it may jump randomly due to the Doppler effect under harsh conditions, forcibly using complex numbers to perform differential will result in garbled text. Degrading to the amplitude (energy) domain can preserve basic diagnostic capabilities (such as distinguishing between electromagnetic leakage and multipath dead zones).

[0066] Therefore, based on the phase correlation between the reference response vector and the next step response vector, either the complex difference mode or the amplitude difference mode is selected.

[0067] As an example, the magnitude of the normalized complex dot product of the aligned lower step response vector and the reference response vector is calculated as the phase correlation coefficient; the phase correlation coefficient is compared with a preset correlation threshold, and either the complex difference mode or the amplitude difference mode is selected.

[0068] In this example, the preset correlation threshold is 0.85. When calculating the phase correlation coefficient, the square of the Euclidean norm of the response vector is used as the total energy of the vector. The product of the Euclidean norm of the aligned next step response vector and the Euclidean norm of the reference response vector is used as the denominator. The complex components of the two vectors at the same index position are multiplied by their conjugates and summed. The absolute value of the sum is used as the numerator, and the ratio of the fractions is used as the phase correlation coefficient. Since the vectors are filtered by the noise floor energy threshold, the energy of the vectors entering the calculation is not zero, and the denominator is not zero. Because time delay alignment has been completed through cross-correlation, the two vectors strictly correspond on the time axis, and each component index has the same physical time delay meaning.

[0069] When the phase correlation coefficient is greater than or equal to the preset correlation threshold, it indicates that the channel is in a coherent steady state. At this time, the Doppler shift or random phase jitter is small, and the angular deviation between the two vectors in the complex plane is within the tolerable range. The complex difference mode can accurately reflect the vector changes caused by physical perturbations (bubbles, displacements), so the complex difference mode is adopted.

[0070] When the phase correlation coefficient is less than the preset correlation threshold, it indicates that the channel is in an incoherent state or a fast fading state. At this time, phase noise dominates, and directly performing complex subtraction will result in an artificially high magnitude of the difference vector. Therefore, it is necessary to degenerate into an amplitude difference mode and only focus on the changes in the energy envelope.

[0071] When the correlation coefficient is 0.85, it corresponds to approximately The average phase rotation error, in underwater acoustic communication engineering, It is usually regarded as the empirical boundary of coherent demodulation, so the preset correlation threshold is set to 0.85 to switch differential modes; in other embodiments of the present invention, the implementer can adjust the preset correlation threshold himself, which will not be described in detail here.

[0072] After selecting the differential mode, the channel change vector can be calculated by comparing the aligned response vector with the selected differential mode.

[0073] As an example, when the complex differential mode is selected, the aligned next step response vector is differentially divided element-wise with the reference response vector to obtain the channel change vector;

[0074] When the amplitude differential mode is selected, the magnitude of each component in the aligned lower step response vector is differentially divided element-wise with the magnitude of each component in the reference response vector to obtain the channel change vector.

[0075] The difference is performed sequentially at the same index position, and the components of the reference response vector are in the subtraction position. The specific vector difference operation process is a conventional technique in the field of signal processing and will not be described in detail.

[0076] Preferably, in one embodiment of the present invention, in order to capture the abnormal energy rise before the main peak of the channel impulse response, the point with the largest modulus of the component in the reference response vector is first extracted as the main peak point. The main peak point refers to the sampling point with the largest energy in the channel impulse response, which is used to define the dominant arrival time of the acoustic wave. In order to avoid the influence of the side lobes of the main peak, the point with a preset guard interval before the main peak point is used as the early arrival cutoff point to obtain the early arrival interval.

[0077] By comparing the energy of the channel change vector within the early arrival interval with the total energy of the channel change vector, the larger the proportion of energy within the early arrival interval to the total energy, the more significant the non-acoustic energy component exists before the arrival of the main path of the sound wave. Since the propagation speed of electromagnetic waves is much higher than that of sound waves, this kind of early arrival energy is very likely to originate from the radio frequency leakage generated by the frequency conversion drive in the umbilical cable. Therefore, this ratio is used as the electromagnetic interference energy ratio and is used as the core basis for judging electromagnetic leakage faults.

[0078] As an example, the index of the main peak is The preset protection interval is 10, corresponding to 10 sampling points. When, the index of the earlier arrival interval is from 0 to In this example, at a sampling rate of 100kHz, the corresponding physical distance is 0.15 meters; when At that time, the electromagnetic interference energy ratio is 0.

[0079] The square of the Euclidean norm of the channel change vector is taken as the total energy of the vector, plus a preset positive parameter divided by zero, such as... Then, the sum of the squares of the magnitudes of each component of the channel change vector within the early arrival interval is used as the denominator, and the sum of the squares of the magnitudes of each component within the early arrival interval is used as the numerator. Finally, the ratio of the fractions is calculated and used as the electromagnetic interference energy ratio.

[0080] It should be noted that, in another embodiment of the present invention, the main peak point can be determined by first performing envelope detection on the reference response vector (e.g., low-pass filtering after modulus taking or smoothing by moving average), and then finding the location of the global maximum value in the smoothed envelope sequence. This method can effectively suppress local peak misjudgment caused by high-frequency noise or phase jitter, improve the robustness of main peak localization, and is especially suitable for high dynamic or low signal-to-noise ratio scenarios.

[0081] In other embodiments of the present invention, the implementer may adjust the size of the preset protection interval according to the actual operating water depth, communication carrier frequency, receiver bandwidth and multipath propagation characteristics. It is advisable to preset the interval through on-site calibration or simulation, which will not be elaborated further.

[0082] Further comparison of the energy difference characteristics between the channel change vector and the reference response vector yields the channel response sensitivity.

[0083] As an example, the square of the Euclidean norm of the channel change vector is used as the total energy of the vector, and this is used as the numerator; the square of the Euclidean norm of the reference response vector is used as the total energy of the vector, plus a preset positive parameter divided by zero, such as... The fractional ratio, used as the denominator, serves as the channel response sensitivity, reflecting the energy difference between the channel change vector and the reference response vector. It characterizes the ratio of the dynamic energy of the perturbation excitation to the total energy of the background channel.

[0084] It should be noted that the mathematical operations involved, such as vector modulus, Euclidean norm, complex conjugate, dot product, and energy normalization, are all techniques well-known to those skilled in the art and will not be elaborated upon further.

[0085] Step S3: Based on the electromagnetic interference energy ratio and channel response sensitivity, identify the communication fault and issue control commands.

[0086] The electromagnetic interference energy ratio reflects whether there is non-acoustic energy that arrives earlier than the sound wave, and the channel response sensitivity reflects whether the channel state changes significantly with the current disturbance of the cleaning pump. Therefore, based on the electromagnetic interference energy ratio and the channel response sensitivity, the logical mutual exclusion classification of fault types can be realized in the early stage of communication degradation, the communication fault can be identified and control commands can be issued, which not only improves the robustness of the system, but also ensures the continuity of deep-sea operations and the safety of equipment.

[0087] Preferably, in one embodiment of the present invention, please refer to Figure 2 The diagram illustrates a flowchart of a method for determining communication faults according to an embodiment of the present invention, specifically including:

[0088] Step S301: When the electromagnetic interference energy ratio is greater than the preset safe energy threshold, it is determined to be an electromagnetic leakage fault.

[0089] As an example, electromagnetic leakage faults correspond to the most serious hardware hazards (such as umbilical cable insulation damage causing high-frequency harmonics of the frequency converter drive to couple to the communication frequency band), which may lead to electrical breakdown or equipment damage. They must be dealt with with the highest priority and are set as the first-level judgment.

[0090] The preset safe energy threshold is 0.5. When the electromagnetic interference energy ratio is greater than the preset safe energy threshold, it indicates that more than half of the dynamically changing energy arrives before the sound wave, indicating the existence of significant non-acoustic pre-arrival energy. Since no real multipath component in the underwater acoustic channel can arrive before the direct sound wave, such a high proportion of pre-arrival energy can only originate from electromagnetic interference propagating along the conductor, and is therefore determined to be an electromagnetic leakage fault.

[0091] Step S302: When the electromagnetic interference energy ratio is less than or equal to the preset safe energy threshold and the channel response sensitivity is less than the preset response correlation threshold, it is determined to be a multipath dead zone fault.

[0092] As an example, the preset response correlation threshold is 0.05. When the electromagnetic interference energy ratio is less than or equal to the preset safe energy threshold, it indicates that the channel degradation is not caused by radio frequency leakage, and hardware insulation failure can be ruled out. At this time, when the channel response sensitivity is less than the preset response correlation threshold, it indicates that the channel state has almost no response to the drastic current adjustment of the cleaning pump. The current communication quality degradation is not related to the cleaning operation itself, but rather to the fact that the robot's spatial position happens to be located at the cancellation point of multipath interference (i.e., the signal dead zone), which is determined to be a multipath dead zone fault.

[0093] This type of fault is determined by the environmental geometry, is location-dependent but poses no risk of equipment damage, and is therefore set as a secondary judgment. Its recovery strategy focuses on location adjustment rather than power cut-off.

[0094] In response to an electromagnetic leakage fault, the controller immediately sends a command to the power management module to disconnect the relay or contactor in the main power supply circuit of the cleaning pump motor. This action is not only to eliminate the radio frequency suppression of communication, but more importantly, to prevent the insulation damage from further expanding under high voltage, thereby protecting the communication module and surrounding sensors from being damaged by breakdown.

[0095] To address multipath dead zone faults, the controller sends commands to the walking actuators, driving the robot to perform an escape and relocation maneuver. Considering the control challenges of deep-sea operations, this embodiment no longer requires micrometer-level precise positioning, but rather controls the robot to move a distance greater than half a wavelength along its current direction of travel or a random direction. The distance (for underwater acoustic communication signals with a carrier frequency of 15kHz, it is approximately greater than 5 centimeters).

[0096] According to the underwater acoustic multipath propagation theory, changing the receiving position by more than half a wavelength is enough to change the superposition phase relationship of the multipath signals, thereby increasing the probability of escaping the current signal interference zero point (dead zone).

[0097] It should be noted that, in another embodiment of the present invention, the "lower step-recovery" of the cleaning pump current constitutes a reversible excitation closed loop. If the channel degradation is caused by fluid effects (such as cavitation bubble clouds), the bubble field will become sparse as the current decreases and reconstruct as the current recovers, resulting in the channel state exhibiting approximately symmetrical but opposite-direction changes during the disturbance and recovery phases.

[0098] Conversely, if the deterioration is caused by rigid body displacement (such as attitude drift caused by recoil force), the system cannot automatically reset in a short time due to mechanical inertia and fluid resistance. The channel change in the recovery phase will deviate significantly from the reverse trajectory of the next step phase. Therefore, when the complex difference mode is selected, it also includes: performing element-wise difference between the aligned next step response vector and the recovery response vector to obtain the recovery change vector.

[0099] Based on the real part of the normalized complex dot product of the channel change vector and the recovery change vector, the change closure index is obtained. The change closure index is used to supplement the judgment of communication faults, thereby distinguishing between fluid reversible faults and mechanical irreversible faults.

[0100] As an example, the two aligned vectors are differencing each other sequentially at the same index position. The components of the aligned next step response vector are in the subtraction position, resulting in the recovery change vector.

[0101] Multiply the complex components of the lower step change vector and the rising change vector at the same index position by conjugate and sum them. Take the real part of the sum and add a negative sign as the numerator. Multiply the Euclidean norm of the lower step change vector and the Euclidean norm of the rising change vector, and add a preset division-by-zero parameter as the denominator. The ratio of the fractions is used as the change closure index.

[0102] In this example, the preset fluid reversibility threshold is 0.7. The supplementary judgment corresponds to the third level of judgment, that is, the judgment is made under the premise that the electromagnetic interference energy ratio is less than or equal to the preset safe energy threshold, and the channel response sensitivity is greater than or equal to the preset response correlation threshold.

[0103] When the change closure index is greater than the preset fluid reversibility threshold, it indicates that the change process is highly reversible and conforms to the characteristics of fluid dynamics, and is judged as a bubble blockage fault.

[0104] When the change closure index is less than or equal to the preset fluid reversibility threshold, it indicates that the change process is irreversible, which conforms to the characteristics of rigid body motion and is judged as a mechanical drift fault.

[0105] To address the bubble blockage fault, the controller sends a command to the cleaning pump driver to reduce the drive current setpoint by a preset percentage (e.g., 20%) by decreasing the PWM duty cycle. Reducing power directly decreases the generation of cavitation nuclei and the density of the bubble cloud, thereby reducing sound wave scattering attenuation. Although cleaning efficiency decreases slightly, the smooth operation and controllability of the communication link are prioritized.

[0106] In response to mechanical drift faults, the controller commands the cleaning pump to temporarily stop (current set to 0) to completely eliminate recoil interference. The system maintains this paused state for a stable period of time (e.g., 5 seconds), utilizing the robot's own gravity and magnetic attraction to naturally adhere to the wall and stabilize its posture without external interference. Once the antenna pointing back to normal, the operation restarts.

[0107] It should be noted that the preset safety energy threshold, preset response correlation threshold, and preset fluid reversibility threshold are not fixed absolute constants; their values ​​can be reasonably configured according to the actual application scenario. In this embodiment, these thresholds can be offline tuned before the equipment leaves the factory through calibration tests in a benchmark environment (e.g., simulating various faults in an anechoic water tank or under known channel conditions).

[0108] In other embodiments of the present invention, when the robot is in a normal, fault-free operating state, the system can continuously collect underlying physical quantities such as steady-state underwater acoustic channel impulse response, received signal noise floor power, and channel energy, and construct a dynamic reference benchmark based on statistical characteristics (such as mean and standard deviation) within a sliding time window. When entering a communication early warning state and executing a fault monitoring process, preset safety energy thresholds, preset response correlation thresholds, etc., can be adaptively adjusted in conjunction with this dynamic benchmark. For example, the absolute early arrival energy threshold of the electromagnetic interference energy ratio can be set to a number of times (such as 3 times) the mean noise floor power of the early arrival zone during the normal period, rather than a fixed ratio, to adapt to changes in background noise levels in different sea areas.

[0109] In summary, to address the technical problem of accurately distinguishing different fault sources based solely on physical layer signal characteristics during communication quality deterioration, this invention provides a method for long-distance communication fault monitoring specifically for underwater cleaning robots. First, under communication warning conditions, the invention controls the robot to execute preset actions and acquire response vectors. Next, using a reference response vector as a benchmark, the lower step response vector and the recovery response vector are time-delay aligned. Then, the channel change vector is calculated by comparing the aligned response vectors. Finally, based on the energy distribution characteristics of the channel change vector before the peak arrives, the electromagnetic interference energy ratio is obtained. Furthermore, based on the energy difference characteristics between the channel change vector and the reference response vector, the channel response sensitivity is obtained. Finally, based on the electromagnetic interference energy ratio and the channel response sensitivity, communication faults are identified and control commands are issued. Under communication warning conditions, this invention excites the channel response through current perturbation, decoupling the fault source based on the energy distribution before the peak and the response sensitivity, achieving accurate identification and self-healing control of faults such as electromagnetic leakage and multipath dead zones, thus solving the technical problem of being unable to distinguish the causes of deterioration.

[0110] It should be noted that the order of the above embodiments of the present invention is merely for descriptive purposes and does not represent the superiority or inferiority of the embodiments. The processes depicted in the accompanying drawings do not necessarily require a specific or sequential order to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0111] The various embodiments in this specification are described in a progressive manner. The same or similar parts between the various embodiments can be referred to each other. Each embodiment focuses on describing the differences from other embodiments.

Claims

1. A method for long-distance communication fault monitoring specifically for underwater cleaning robots, characterized in that, The method includes: In the communication warning state, the load driving current of the robot is controlled to perform preset steady-state holding, lower step perturbation and recovery actions in sequence, and the underwater acoustic channel impulse response at each stage is obtained, which are the reference response vector, lower step response vector and recovery response vector, respectively. Using the reference response vector as a benchmark, the lower step response vector and the recovery response vector are time-delay aligned respectively; the channel change vector is calculated by comparing the aligned response vectors; the electromagnetic interference energy ratio is obtained based on the energy distribution characteristics of the channel change vector before the arrival of the main peak; and the channel response sensitivity is obtained based on the energy difference characteristics between the channel change vector and the reference response vector. Based on the electromagnetic interference energy ratio and channel response sensitivity, communication faults are identified and control commands are issued. The methods for identifying communication faults include: when the electromagnetic interference energy ratio is greater than a preset safe energy threshold, it is determined to be an electromagnetic leakage fault; when the electromagnetic interference energy ratio is less than or equal to the preset safe energy threshold, and the channel response sensitivity is less than a preset response correlation threshold, it is determined to be a multipath dead zone fault. The method for obtaining the channel change vector includes: selecting either the complex differential mode or the amplitude differential mode based on the phase correlation between the reference response vector and the next step response vector; and calculating the channel change vector by comparing the aligned response vectors based on the selected differential mode. When the complex difference mode is selected, the method further includes: performing element-wise difference between the aligned lower step response vector and the recovery response vector to obtain the recovery change vector; obtaining the change closure index based on the real part of the normalized complex dot product of the channel change vector and the recovery change vector; and supplementing the judgment of communication faults based on the change closure index; when the change closure index is greater than the preset fluid reversibility threshold, it is judged as a bubble blockage fault; when the change closure index is less than or equal to the preset fluid reversibility threshold, it is judged as a mechanical drift fault.

2. The method for long-distance communication fault monitoring of underwater cleaning robots according to claim 1, characterized in that, The method for selecting the difference mode includes: The magnitude of the normalized complex dot product of the aligned lower step response vector and the reference response vector is calculated and used as the phase correlation coefficient. The phase correlation coefficient is compared with a preset correlation threshold, and either the complex difference mode or the amplitude difference mode is selected.

3. The method for long-distance communication fault monitoring of underwater cleaning robots according to claim 1, characterized in that, When the complex differential mode is selected, the aligned lower step response vector is differentially divided element-wise with the reference response vector to obtain the channel change vector; When the amplitude differential mode is selected, the magnitude of each component in the aligned lower step response vector is differentially divided element-wise with the magnitude of each component in the reference response vector to obtain the channel change vector.

4. A method for long-distance communication fault monitoring for underwater cleaning robots according to claim 1, characterized in that, The method for obtaining the electromagnetic interference energy ratio includes: The point with the largest modulus of the component in the reference response vector is extracted as the main peak point, and the point with the preset protection interval before the main peak point is taken as the early arrival cutoff point to obtain the early arrival interval. The electromagnetic interference energy ratio is obtained by comparing the energy of the channel change vector within the early arrival interval with the total energy of the channel change vector.

5. A method for long-distance communication fault monitoring for underwater cleaning robots according to claim 1, characterized in that, The control commands include: In response to the electromagnetic leakage fault, the controller immediately sends a command to the power management module to disconnect the relay or contactor of the main power supply circuit of the cleaning pump motor; In response to the multipath dead zone fault, the controller sends a command to the walking driver to drive the robot to perform an escape and relocation action.

Images