阅读说明：本技术 一种搜寻水下失踪者的rov探测系统及探测方法 (ROV detection system and detection method for searching underwater missing person ) 是由 盛明伟 王玮哲 秦洪德 朱仲本 万磊 褚玲怡 李俊 于 2020-06-23 设计创作，主要内容包括：本发明提供一种搜寻水下失踪者的ROV探测系统及探测方法,水下工作部分包括水下主控制系统、水下视觉探测系统、机械臂、水下发声装置、水下声音收集装置。水下主控制系统用来与母船进行信息交互,同时对水下各作业部分进行整体控制。机械臂A用于连接ROV与沉船,同时机械臂A搭载水下发声装置和水下声音接收装置。水下发声装置用于向沉船内失踪人员进行广播。水下声音收集装置用于收集失踪人员反馈的敲击声音和沉船附近的声音,并实时上传到母船上的声音在线处理系统进行处理。机械臂B用于敲击沉船。母船工作部分包括母船主控制系统、视觉观察控制系统和声音在线处理系统。本发明提高救援效率和准确性。(The invention provides an ROV detection system and a detection method for searching an underwater lost person. The underwater main control system is used for carrying out information interaction with the mother ship and simultaneously carrying out overall control on all underwater operation parts. The mechanical arm A is used for connecting the ROV and the sunken ship, and meanwhile, the mechanical arm A carries the underwater sound generating device and the underwater sound receiving device. The underwater sound production device is used for broadcasting to the missing person in the sunken ship. The underwater sound collecting device is used for collecting knocking sound fed back by the missing person and sound near the sunken ship and uploading the knocking sound and the sound near the sunken ship to the sound online processing system on the mother ship in real time for processing. The mechanical arm B is used for knocking the sunken ship. The mother ship working part comprises a mother ship main control system, a visual observation control system and a sound online processing system. The invention improves the rescue efficiency and accuracy.)

1. An ROV detection system for searching for an underwater missing person, comprising: the underwater working part comprises an underwater main control system, an underwater vision detection system, a mechanical arm, an underwater sound production device and an underwater sound collection device; the mother ship working part comprises a mother ship main control system, a visual observation control system and a sound online processing system; the underwater working part is connected with the mother ship working part through an umbilical cable, video information collected by the underwater working part is uploaded to the mother ship visual observation control system in real time, and sound data collected by the underwater working part is uploaded to the mother ship sound online processing system in real time.

2. An ROV detection system for searching for an underwater missing person according to claim 1 wherein: the underwater main control system is used for carrying out information interaction with the mother ship, simultaneously carrying out integral control on all underwater operation parts and sending underwater operation instructions; the underwater vision detection system comprises an underwater high-definition network camera and an underwater LED lamp, wherein the underwater high-definition network camera is used for shooting an underwater environment and transmitting a video stream to a master ship main control system through an optical transmitter and receiver, so that personnel on a master ship can observe an ROV operation environment and the normal working state of a mechanical arm in real time, and the underwater LED lamp is used for providing an auxiliary light source; the tail part of the mechanical arm A is provided with an adsorption device for connecting the ROV and the sunken ship, and meanwhile, the mechanical arm A carries an underwater sound production device and an underwater sound receiving device; the mechanical arm B adopts a swing arm type knocking mode or carries a knocking device at the tail part of the mechanical arm B to knock the sunken ship; the underwater sound generating device is used for broadcasting to the missing person in the sunken ship, guiding the missing person to respond in a mode of regularly knocking the ship body after the mechanical arm knocks, and matching with rescue actions; the underwater sound collecting device is used for collecting knocking sound fed back by the missing person and sound near the sunken ship, and audio is uploaded to a sound online processing system on the mother ship in real time through a network switch and an optical transceiver for processing.

3. An ROV detection system for searching for an underwater missing person according to claim 2 wherein: the network switch is an equipment for expanding a network, and can provide more connecting ports for a sub-network so as to connect an underwater main control system, an underwater vision detection system, an underwater sound production device and an underwater sound collection device.

4. An ROV detection system for searching for a missing person underwater according to claim 1, 2 or 3 wherein: the master ship main control system is used for receiving information transmitted by the ROV and sending instructions to the ROV; the visual observation control system comprises an operation table and a display, and is used for displaying the underwater condition in real time, so that a worker on a mother ship can monitor the underwater condition in real time and remotely control the ROV; the processing method of the sound online processing system comprises the steps of matching a computer deep learning model and manually carrying out online monitoring and identification, and carrying out man-machine cooperative identification on a sound file.

5. An ROV detection system for searching for an underwater missing person according to claim 4 wherein: the computer in the sound online processing system adopts a deep learning method for recognizing and detecting the sound: the method comprises the steps of collecting sounds of knocking different hull materials under various conditions in advance, training a deep learning model for sound recognition, matching and comparing the collected sound wave characteristics by using the model in a sound online processing system during rescue, marking and storing a high-confidence position of the knocking sound of a suspected missing person, carrying out integral calculation to obtain a final result gamma by combining the manual monitoring result of staff on a mother ship on sound information, setting the weight of the result calculated by a computer as alpha and the weight of the result calculated by the computer as beta, and setting alpha + beta as 1 at the same time, and setting a threshold value, wherein if gamma is larger than or equal to the threshold value, the missing person is judged to exist; if γ is less than the threshold, then no missing person is present.

6. An ROV detection method for searching an underwater missing person, characterized in that: according to the ROV detection system for searching the lost person underwater, the steps are as follows:

the method comprises the following steps: the underwater main control system sends a power-on command to electrify equipment of each part such as the underwater vision detection system, the mechanical arm, the underwater sound production device and the underwater sound collection device;

step two: after arriving at the area near the sunken ship, a worker on the mother ship observes a real-time picture shot by a high-definition network camera, operates an ROV to approach the sunken ship, controls a mechanical arm A to approach the hull of the sunken ship, starts an adsorption device and fixes the mechanical arm A on the hull of the sunken ship; starting an underwater sound receiving device, collecting sound near a sunken ship, starting an underwater sound generating device to broadcast for 30 seconds to persons in distress in the sunken ship, wherein the broadcast content is to guide the missing person to respond in a mode of regularly knocking a ship body after knocking by a mechanical arm B; waiting for half a minute after broadcasting is finished, starting the mechanical arm B to regularly knock the sunken ship body, waiting for several minutes in situ after knocking is finished, and keeping the underwater sound receiving device started in the whole process and continuously collecting sound information in a nearby range;

step three: uploading the collected sound information to a sound online processing system of the mother ship in real time, and carrying out man-machine cooperative detection; a computer in the sound online processing system analyzes and compares sound in real time by using a trained deep learning model, judges whether the sound is a missing person according to the confidence of a detection result, and marks and stores a suspicious position; workers on the mother ship simultaneously carry out manual monitoring on underwater sound, judge whether artificial knocking sound exists or not and record results; combining the two detection results, and if the artificial knocking information is collected, performing a fourth step; if the artificial knocking information is not collected, repeating the step two once, and if the artificial knocking information is not collected, displaying and then performing the step four;

step four: after the sound on-line processing system on the mother ship detects, closing the adsorption device, disconnecting the mechanical arm from the hull of the sunken ship, operating the ROV by a worker to move to other positions of the sunken ship in a clockwise or anticlockwise sequence, and repeating the second step and the third step until the sound on-line processing system stops after surrounding the hull for one circle; recording the detected suspicious position with high confidence level so as to break the position point in further rescue;

step five: and (5) ending the task and recycling the ROV.

Technical Field

The invention relates to an ROV detection system and a detection method for searching an underwater missing person, and belongs to the field of underwater detection and rescue.

Background

The surface of the earth has about 71 percent of water, and the coverage area of the water is far larger than the land area, so that the reasonable development and utilization of water resources are the most practical and potential resource utilization mode. When water resources are utilized, people can easily find that various ships are often used as carriers in both overwater operation and underwater operation. With the continuous development of economy and technology, shipbuilding technology is more mature, but mechanical failures and natural disasters can also cause the occurrence of ship sinking accidents. The sudden nature of the accident may make some people have no time to escape, and the people are trapped inside the ship and sink to the water bottom together with the ship, and in order to confirm the information such as the specific positions of the lost people, rescue workers are often required to go to the bottom of the water area of the accident for manual detection. The rescue personnel need too much equipment to carry, and the rescue personnel often face unknown underwater risks such as severe environment in accident areas, and the problems can influence the progress of rescue and are not beneficial to rescue. In order to guarantee the safety of rescue personnel, improve the rescue efficiency and accelerate the rescue progress and accuracy, the unmanned remote control submersible can be used for rescue.

The Remote Operated Vehicle (ROV) has the advantages of flexible operation, good economy and the like, can finish the observation tasks of ports and docks, ocean structures, submarine topography and the like, can carry relatively complete equipment to detect in real time, can replace divers to enter dangerous environments or deep water environments to work, and has long-term application prospects in various fields. For example, 4500-meter deep sea operation system "sea horse" number smoothly carries out comprehensive sea test experiment and delivers in 2014 in south sea, can accomplish submarine operation tasks such as submarine pipeline maintenance, arrangement and recovery of submarine observation network equipment instruments and the like. The invention provides an ROV detection system for searching an underwater missing person based on sound feedback, which can realize the detection of the trapped missing person by a mode of combining deep learning and manual identification after sound is collected.

Disclosure of Invention

The invention provides an ROV detection system and a detection method suitable for searching an underwater missing person by an underwater robot in disaster areas such as oceans, lakes, rivers and the like based on sound feedback.

The purpose of the invention is realized as follows: the underwater working part comprises an underwater main control system, an underwater vision detection system, a mechanical arm, an underwater sound production device and an underwater sound collection device; the mother ship working part comprises a mother ship main control system, a visual observation control system and a sound online processing system; the underwater working part is connected with the mother ship working part through an umbilical cable, video information collected by the underwater working part is uploaded to the mother ship visual observation control system in real time, and sound data collected by the underwater working part is uploaded to the mother ship sound online processing system in real time.

The invention also includes such structural features:

1. the underwater main control system is used for carrying out information interaction with the mother ship, simultaneously carrying out integral control on all underwater operation parts and sending underwater operation instructions; the underwater vision detection system comprises an underwater high-definition network camera and an underwater LED lamp, wherein the underwater high-definition network camera is used for shooting an underwater environment and transmitting a video stream to a master ship main control system through an optical transmitter and receiver, so that personnel on a master ship can observe an ROV operation environment and the normal working state of a mechanical arm in real time, and the underwater LED lamp is used for providing an auxiliary light source; the tail part of the mechanical arm A is provided with an adsorption device for connecting the ROV and the sunken ship, and meanwhile, the mechanical arm A carries an underwater sound production device and an underwater sound receiving device; the mechanical arm B adopts a swing arm type knocking mode or carries a knocking device at the tail part of the mechanical arm B to knock the sunken ship; the underwater sound generating device is used for broadcasting to the missing person in the sunken ship, guiding the missing person to respond in a mode of regularly knocking the ship body after the mechanical arm knocks, and matching with rescue actions; the underwater sound collecting device is used for collecting knocking sound fed back by the missing person and sound near the sunken ship, and audio is uploaded to a sound online processing system on the mother ship in real time through a network switch and an optical transceiver for processing.

2. The network switch is an equipment for expanding a network, and can provide more connecting ports for a sub-network so as to connect an underwater main control system, an underwater vision detection system, an underwater sound production device and an underwater sound collection device.

3. The master ship main control system is used for receiving information transmitted by the ROV and sending instructions to the ROV; the visual observation control system comprises an operation table and a display, and is used for displaying the underwater condition in real time, so that a worker on a mother ship can monitor the underwater condition in real time and remotely control the ROV; the processing method of the sound online processing system comprises the steps of matching a computer deep learning model and manually carrying out online monitoring and identification, and carrying out man-machine cooperative identification on a sound file.

4. The computer in the sound online processing system adopts a deep learning method for recognizing and detecting the sound: the method comprises the steps of collecting sounds of knocking different hull materials under various conditions in advance, training a deep learning model for sound recognition, matching and comparing the collected sound wave characteristics by using the model in a sound online processing system during rescue, marking and storing a high-confidence position of the knocking sound of a suspected missing person, carrying out integral calculation to obtain a final result gamma by combining the manual monitoring result of staff on a mother ship on sound information, setting the weight of the result calculated by a computer as alpha and the weight of the result calculated by the computer as beta, and setting alpha + beta as 1 at the same time, and setting a threshold value, wherein if gamma is larger than or equal to the threshold value, the missing person is judged to exist; if γ is less than the threshold, then no missing person is present.

5. An ROV detection method for searching an underwater missing person, which is based on an ROV detection system for searching the underwater missing person, and comprises the following steps:

the method comprises the following steps: the underwater main control system sends a power-on command to electrify equipment of each part such as the underwater vision detection system, the mechanical arm, the underwater sound production device and the underwater sound collection device;

step two: after arriving at the area near the sunken ship, a worker on the mother ship observes a real-time picture shot by a high-definition network camera, operates an ROV to approach the sunken ship, controls a mechanical arm A to approach the hull of the sunken ship, starts an adsorption device and fixes the mechanical arm A on the hull of the sunken ship; starting an underwater sound receiving device, collecting sound near a sunken ship, starting an underwater sound generating device to broadcast for 30 seconds to persons in distress in the sunken ship, wherein the broadcast content is to guide the missing person to respond in a mode of regularly knocking a ship body after knocking by a mechanical arm B; waiting for half a minute after broadcasting is finished, starting the mechanical arm B to regularly knock the sunken ship body, waiting for several minutes in situ after knocking is finished, and keeping the underwater sound receiving device started in the whole process and continuously collecting sound information in a nearby range;

step three: uploading the collected sound information to a sound online processing system of the mother ship in real time, and carrying out man-machine cooperative detection; a computer in the sound online processing system analyzes and compares sound in real time by using a trained deep learning model, judges whether the sound is a missing person according to the confidence of a detection result, and marks and stores a suspicious position; workers on the mother ship simultaneously carry out manual monitoring on underwater sound, judge whether artificial knocking sound exists or not and record results; combining the two detection results, and if the artificial knocking information is collected, performing a fourth step; if the artificial knocking information is not collected, repeating the step two once, and if the artificial knocking information is not collected, displaying and then performing the step four;

step four: after the sound on-line processing system on the mother ship detects, closing the adsorption device, disconnecting the mechanical arm from the hull of the sunken ship, operating the ROV by a worker to move to other positions of the sunken ship in a clockwise or anticlockwise sequence, and repeating the second step and the third step until the sound on-line processing system stops after surrounding the hull for one circle; recording the detected suspicious position with high confidence level so as to break the position point in further rescue;

step five: and (5) ending the task and recycling the ROV.

Compared with the prior art, the invention has the beneficial effects that: the ROV detection system for searching the underwater missing person based on the sound feedback uses the adsorption device on the mechanical arm to connect the ROV and the sunken ship, and ensures that the ROV can keep the relative position with the sunken ship in severe environments such as waves and the like or during the self arm swinging action. The ROV detection system for searching the underwater missing person based on the sound feedback adopts a deep learning method, is matched with manual detection, calculates corresponding results by using the occupied weight, and improves the rescue efficiency and accuracy. The ROV detection system for searching the underwater missing person based on the sound feedback can simultaneously use a multi-ROV cooperative detection mode, and arrange a plurality of ROVs to surround the sunken ship for detection, thereby further shortening the rescue time.

Due to the fact that most underwater environments are severe in the water area where the wrecker sinks, in order to prevent the ROV from generating large deviation of the relative position of the ROV and the sunken ship due to potential undercurrent or swinging during operation, the ROV is connected with the hull of the sunken ship through a mechanical arm. Meanwhile, an underwater sound collecting device can be additionally arranged on the ROV body to monitor the sound near the sunken ship, so that valuable sound information is prevented from being omitted. For the ROV operation position when suspected trapped people are found, equipment such as a depth meter, a DVL, a compass, an inertial navigation device and the like carried by the ROV can be used for calculating and obtaining through position-push navigation, and the position of the ROV is assisted and positioned in combination with USBL ultra-short baseline positioning, so that the positioning precision of the suspected lost people is improved and recorded, and a basis is provided for the position of the next rescue breaking operation. In order to improve efficiency and shorten search detection time, the invention can simultaneously use a multi-ROV cooperative detection mode to arrange a plurality of ROVs to surround a sunken ship for detection, carries out on-line fusion processing on sound signals uploaded by the plurality of ROVs, carries out sequencing by utilizing the confidence level of detection results and confirms the position of a missing person as soon as possible.

Drawings

Fig. 1 is a schematic diagram of an ROV system.

Fig. 2 is a flow chart of the ROV operation mode.

Detailed Description

The present invention will be described in further detail with reference to the accompanying drawings and specific embodiments.

As shown in fig. 1, an ROV detection system for searching for an underwater missing person based on acoustic feedback includes an underwater working part and a mother ship working part. The underwater working part comprises an underwater main control system, an underwater vision detection system, a mechanical arm, an underwater sound production device and an underwater sound collection device. The underwater main control system is used for carrying out information interaction with the mother ship, simultaneously carrying out integral control on all underwater operation parts and sending underwater operation instructions. The underwater vision detection system comprises an underwater high-definition network camera and an underwater LED lamp. The underwater high-definition network camera is used for shooting an underwater environment, transmitting video streams to the master control system of the mother ship through the optical transceiver, enabling workers on the mother ship to observe the ROV operation environment and the normal working state of the mechanical arm in real time, and the underwater LED lamp is used for providing an auxiliary light source. The underwater sound generating device comprises two mechanical arms, an adsorption device is mounted at the tail of the mechanical arm A and used for connecting the ROV and the sunken ship, and meanwhile the mechanical arm A carries an underwater sound generating device and an underwater sound receiving device. The underwater sound generating device is used for broadcasting to the missing person in the sunken ship, guiding the missing person to respond in a mode of regularly knocking the ship body after the mechanical arm knocks, and matching with rescue actions. The underwater sound collecting device is used for collecting knocking sound fed back by the missing person and sound near the sunken ship, and audio is uploaded to a sound online processing system on the mother ship in real time through a network switch and an optical transceiver for processing. The mechanical arm B adopts a swing arm type knocking mode or carries a knocking device at the tail part of the mechanical arm B to knock the sunken ship. The network switch is an equipment for expanding a network, and can provide more connecting ports for a sub-network so as to be connected with an underwater main control system, an underwater vision detection system, an underwater sound production device and an underwater sound collection device. The mother ship working part comprises a mother ship main control system, a visual observation control system and a sound online processing system. And the mother ship main control system is used for receiving the information transmitted by the ROV and sending an instruction to the ROV. The visual observation control system comprises an operation table and a display, and is used for displaying the underwater condition in real time, so that the staff on the mother ship can monitor the underwater condition in real time and remotely control the ROV. The processing method of the sound online processing system comprises the steps of matching a computer deep learning model and manually monitoring and identifying online, man-machine collaborative identification of sound files is carried out, and the efficiency and the accuracy of judgment are improved. The underwater working part and the mother ship working part in the detection system are connected through an umbilical cable, video information collected by the underwater working part is uploaded to the mother ship visual observation control system in real time, and sound data collected by the underwater working part is uploaded to the mother ship sound online processing system in real time.

As shown in fig. 2, an ROV detection system for searching for an underwater missing person based on sound feedback works as follows:

the method comprises the following steps: and the underwater main control system sends a power-on command to electrify equipment of the underwater vision detection system, the mechanical arm, the underwater sound production device, the underwater sound collection device and the like.

Step two: after the ship arrives at the area near the sunken ship, workers on the mother ship observe real-time pictures shot by a high-definition network camera, operate the ROV to approach the sunken ship, control the mechanical arm A to approach the hull of the sunken ship, start the adsorption device and fix the mechanical arm A on the hull of the sunken ship. And starting the underwater sound receiving device to collect the sound near the sunken ship, starting the underwater sound generating device to broadcast for 30 seconds to the persons in distress in the sunken ship, wherein the broadcast content is used for guiding the missing person to respond in a mode of regularly knocking the ship body after knocking by the mechanical arm B. And after the broadcast is finished, waiting for half a minute, starting the mechanical arm B to regularly knock the sunken ship body, after knocking, waiting for several minutes in situ, and keeping the underwater sound receiving device started in the whole process and continuously collecting sound information in a nearby range.

Step three: and uploading the collected sound information to a sound online processing system of the mother ship in real time to carry out man-machine cooperative detection. And a computer in the sound online processing system analyzes and compares the sound in real time by using the trained deep learning model, judges whether the person is a missing person according to the confidence coefficient of the detection result, and marks and stores the suspicious position. And simultaneously, the workers on the mother ship monitor underwater sound manually, judge whether the artificial knocking sound exists or not and record the result. And combining the two detection results, repeating the step two again if the artificial knocking information is not collected, and turning to the step four if the artificial knocking information is not collected.

Step four: in order to carry out complete detection on the whole sunken ship, after the sound on-line processing system on the mother ship detects, the adsorption device is closed, the mechanical arm is disconnected with the sunken ship body, a worker controls the ROV to move to other positions of the sunken ship body in a clockwise or anticlockwise sequence, and the second step and the third step are repeated until the sound on-line processing system stops after surrounding the ship body for one circle. And recording the detected suspicious position with high confidence so as to break down the position point in further rescue.

Step five: and (5) ending the task and recycling the ROV.

The ROV detection system for searching the underwater missing person based on the sound feedback has the advantages that due to the fact that the lost ship is sunk in a water area, most underwater environment conditions are severe, and in order to prevent the ROV from generating large deviation with the relative position of a sunken ship when the ROV swings due to possible undercurrent or self swinging in the operation process, the ROV is connected with the ship body of the sunken ship through the mechanical arm. Meanwhile, an underwater sound collecting device can be additionally arranged on the ROV body to monitor the sound near the sunken ship, so that valuable sound information is prevented from being omitted.

The ROV detection system for searching the underwater missing person based on the sound feedback is characterized in that a deep learning method is adopted for recognizing and detecting the sound by a computer in the sound online processing system. The method comprises the steps of collecting sounds of knocking different hull materials under various conditions in advance, training a deep learning model for sound recognition, matching and comparing the collected sound wave characteristics by using the model in a sound online processing system during rescue, marking and storing a high-confidence position of the knocking sound of a suspected missing person, carrying out integral calculation by combining the manual monitoring result of staff on a mother ship on sound information, setting the weight of the calculation result of a computer as alpha and the weight of the manual monitoring result as beta, and setting alpha + beta as 1 at the same time to obtain a final result gamma, and setting a threshold value, wherein if gamma is larger than the value, the missing person exists. Therefore, man-machine cooperative operation is realized, and the judgment efficiency and accuracy are improved.

The ROV detection system for searching the underwater missing person based on the sound feedback can be used for calculating the ROV operation position when the suspected trapped person is found by using equipment such as a depth meter, a DVL, a compass and an inertial navigation device carried by the ROV through position-push navigation, and can be used for carrying out auxiliary positioning on the ROV position by combining USBL ultra-short baseline positioning, so that the positioning precision of the suspected missing person is improved and recorded, and a basis is provided for the position of the next rescue breaking-in operation.

The ROV detection system for searching the underwater missing person based on the sound feedback can be used for improving the efficiency and shortening the searching and detecting time, a multi-ROV cooperative detection mode can be used at the same time, a plurality of ROVs are distributed to surround a sunken ship for detection, sound signals uploaded by the plurality of ROVs are subjected to fusion processing on line, sequencing is carried out by using the confidence degree of detection results, and the position of the missing person is confirmed as soon as possible.

In summary, the invention relates to the field of underwater detection and rescue, in particular to an ROV detection system and a corresponding detection method thereof, which are suitable for searching an underwater missing person in disaster areas such as oceans, lakes, rivers and the like by ROV based on sound feedback. The underwater working part comprises an underwater main control system, an underwater vision detection system, a mechanical arm, an underwater sound production device and an underwater sound collection device. The underwater main control system is used for carrying out information interaction with the mother ship and simultaneously carrying out overall control on all underwater operation parts. The underwater vision detection system comprises an underwater high-definition network camera and an underwater LED lamp and is used for shooting an underwater environment and uploading video streams to the master ship main control system in real time. The total two mechanical arms are used for connecting the ROV and the sunken ship, and meanwhile, the underwater sound generating device and the underwater sound receiving device are carried on the mechanical arms. The underwater sound production device is used for broadcasting to the missing person in the sunken ship. The underwater sound collecting device is used for collecting knocking sound fed back by the missing person and sound near the sunken ship and uploading the knocking sound and the sound near the sunken ship to the sound online processing system on the mother ship in real time for processing. The mechanical arm B is used for knocking the sunken ship. The mother ship working part comprises a mother ship main control system, a visual observation control system and a sound online processing system. And the master ship main control system is used for receiving the information transmitted by the ROV and sending instructions to the ROV. The visual observation control system comprises an operation table and a display, so that workers on the mother ship can monitor underwater conditions in real time and remotely control the ROV. The processing method of the sound online processing system comprises the steps of matching a computer deep learning model and manually monitoring and identifying online, man-machine collaborative identification of sound files is carried out, and the efficiency and the accuracy of judgment are improved.