Nationwide architecture sonar analysis and fault early warning method, system and device and medium

Abstract

The invention relates to the field of artificial intelligence computing technology, and provides a nationwide architecture sonar analysis and fault early warning method, system and device and a medium, and the method comprises the steps: obtaining real-time data, and carrying out the reasoning of the real-time data, and obtaining an observation result; inputting the real-time data and the observation result into a preset fault prediction model to obtain a first prediction result; comparing the first prediction result with the real-time data to obtain fault features, and performing clustering analysis on the fault features to obtain a second prediction result; and performing fault alarm based on the second prediction result. On one hand, an underwater target can be monitored in real time, on the other hand, fault prediction can be carried out when the system is about to be abnormal, early warning is carried out in advance once the system is found, an alarm is automatically given in real time, monitoring personnel are reminded to carry out maintenance in time, real-time monitoring, target object recognition and fault prediction on the sonar system are achieved, and the system is high in practicability. And the performance and the safety of the sonar system can be obviously improved.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence computing technology, and in particular to methods, systems, equipment and media for sonar analysis and fault warning of a domestically produced architecture. Background Art

[0002] As one of the most widely used technologies in hydroacoustics, the performance of sonar is directly related to the accuracy of underwater target detection, positioning and communication.

[0003] With the in-depth implementation of the marine strategy, sonar technology plays an increasingly important role in marine detection, underwater target identification, marine environment monitoring, etc. However, traditional sonar data analysis and fault prediction methods mostly rely on imported hardware and software, which poses data security risks and technical control problems.

[0004] In addition, the existing sonar data analysis and fault prediction systems have problems such as strong technical dependence, poor data security, and low prediction accuracy due to the reliance on simple threshold judgment and statistical methods. In actual applications, sonar systems face problems such as complex underwater acoustic environments and equipment aging and performance degradation caused by long-term operation. However, existing sonar systems rely on imported technology and equipment, which poses data security risks and is costly. Summary of the invention

[0005] In order to solve the above problems, the present invention realizes the comprehensive localization of hardware and software, improves data security, and enhances the accuracy of fault prediction.

[0006] Based on the above objectives, the present invention proposes a method for sonar analysis and fault warning of a domestically produced architecture, including: Acquire real-time data, and perform reasoning on the real-time data to obtain observation results; Inputting the real-time data and the observation result into a preset fault prediction model to obtain a first prediction result; Comparing the first prediction result with the real-time data to obtain fault characteristics, and performing cluster analysis on the fault characteristics to obtain a second prediction result; A fault alarm is issued based on the second prediction result.

[0007] In some embodiments, the real-time data includes one or more of multi-channel sonar signals and system status data; wherein the system status data includes one or more of voltage, current and temperature.

[0008] In some embodiments, the step of inferring the real-time data to obtain observation results includes: Performing signal conditioning and analog-to-digital conversion on the multi-channel sonar signals in the real-time data; The converted signal is stored and preprocessed to obtain key features; Real-time reasoning is performed on the key features to obtain observation results.

[0009] In some embodiments, the steps of constructing the preset fault prediction model are as follows: Acquire a historical data set, and process the historical data set to obtain a feature data set; Through the LSTM model, the mean square error is selected as the loss function, and the RMSprop optimizer is used to train with the feature data set to obtain a fault prediction model.

[0010] In some embodiments, the step of comparing the first prediction result with the real-time data to obtain a fault feature includes: Comparing the first prediction result with the real-time data and calculating the residual; According to a preset residual distribution threshold, residuals exceeding the preset threshold are regarded as fault features.

[0011] In some embodiments, the step of performing cluster analysis on the fault features to obtain a second prediction result includes: Cluster the fault features, classify similar fault features into the same category, and obtain clustering results; The clustering result is input into a prediction model based on clustering to perform prediction, thereby obtaining a second prediction result.

[0012] In some embodiments, the step of issuing a fault warning based on the second prediction result includes: In response to the second prediction result being a fault caused by the target object, directly triggering an alarm; In response to the second prediction result being that the fault is not caused by the target object, the fault type is reported, and maintenance is performed based on the fault type.

[0013] The present invention proposes a sonar analysis and fault warning system with a nationally produced architecture, comprising: An acquisition unit, configured to acquire real-time data and perform reasoning on the real-time data to obtain an observation result; A prediction unit, configured to input the real-time data and the observation result into a preset fault prediction model to obtain a first prediction result; A clustering unit, configured to compare the first prediction result with the real-time data to obtain a fault feature, and perform cluster analysis on the fault feature to obtain a second prediction result; An alarm unit is configured to issue a fault alarm based on the second prediction result.

[0014] The present invention provides a computer device, comprising: At least one processor; and a memory, wherein the memory stores a computer program that can be run on the processor, and when the processor executes the program, the steps of the method for sonar analysis and fault warning of the national domestic architecture are performed.

[0015] The present invention proposes a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the method for sonar analysis and fault warning of a nationally produced architecture are executed.

[0016] The present invention has at least the following beneficial technical effects: The present invention proposes a method, system, device and medium for sonar analysis and fault warning of a nationally produced architecture, the method comprising: Acquire real-time data, perform inference on the real-time data to obtain observation results; input the real-time data and the observation results into a preset fault prediction model to obtain a first prediction result; compare the first prediction result with the real-time data to obtain fault characteristics, perform cluster analysis on the fault characteristics to obtain a second prediction result; and issue a fault alarm based on the second prediction result.

[0017] On the one hand, the present invention can monitor underwater targets in real time, and on the other hand, it can predict faults when the system is about to become abnormal. Once an abnormality is found, it can give an early warning and automatically issue an alarm in real time to remind monitoring personnel to perform maintenance in time, thereby realizing real-time monitoring, target identification and fault prediction of the sonar system. It is efficient, stable and reliable, and can significantly improve the performance and safety of the sonar system. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention. For ordinary technicians in this field, other embodiments can be obtained based on these drawings without paying creative work.

[0019] Figure 1 A flow chart of the method for sonar analysis and fault warning of a nationally produced architecture provided by the present invention; Figure 2 A system module diagram of sonar analysis and fault warning for a nationally produced architecture provided by the present invention; Figure 3 An architecture diagram of an embodiment of a sonar analysis and fault warning system with a domestically-produced architecture provided by the present invention; Figure 4A flow chart of an embodiment of a method for sonar analysis and fault warning of a domestically produced architecture provided by the present invention; Figure 5 A schematic diagram of the structure of an embodiment of a computer device provided by the present invention; Figure 6 A schematic diagram of the structure of an embodiment of a computer-readable storage medium provided by the present invention. DETAILED DESCRIPTION

[0020] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the embodiments of the present invention are further described in detail below in combination with specific embodiments and with reference to the accompanying drawings.

[0021] It should be noted that all expressions using "first" and "second" in the embodiments of the present invention are for distinguishing two non-identical entities with the same name or non-identical parameters. It can be seen that "first" and "second" are only for the convenience of expression and should not be understood as limitations on the embodiments of the present invention. The subsequent embodiments will not explain this one by one.

[0022] The present invention proposes a method for sonar analysis and fault warning of a domestically produced architecture. Figure 1 ,include: S1: Acquire real-time data, and perform reasoning on the real-time data to obtain observation results; S2: inputting the real-time data and the observation result into a preset fault prediction model to obtain a first prediction result; S3: Compare the first prediction result with the real-time data to obtain fault features, and perform cluster analysis on the fault features to obtain a second prediction result; S4: Issue a fault alarm based on the second prediction result.

[0023] Using advanced sound analysis technology, sonar data is processed and analyzed in real time to detect potential faults and send out early warning signals in a timely manner. Once a fault or abnormal situation is detected, the system can respond quickly and take measures to deal with it to avoid the expansion of the fault or serious consequences. The domestic architecture means that everything from hardware to software has been localized, which greatly reduces dependence on external technology and supply chains and enhances independent control capabilities in key technologies and equipment.

[0024] In some embodiments, see Figure 1 , the real-time data includes one or more of multi-channel sonar signals and system status data; wherein the system status data includes one or more of voltage, current and temperature.

[0025] By monitoring multiple sonar signals simultaneously, more comprehensive information about the seabed topography, target distribution, and underwater acoustic environment can be obtained. This helps reduce the errors that may be caused by a single signal and improves the precision and accuracy of monitoring. Real-time monitoring of system status data, such as the working status of the sensor and the load of the signal processing unit, can promptly detect and eliminate potential fault factors and ensure the stable operation of the sonar system, thereby further improving the accuracy of monitoring data. It helps to ensure the continuous and stable operation of the sonar system in complex environments.

[0026] In some embodiments, see Figure 1 , the step of inferring the real-time data to obtain the observation result comprises: Performing signal conditioning and analog-to-digital conversion on the multi-channel sonar signals in the real-time data; The converted signal is stored and preprocessed to obtain key features; The key features are inferred in real time to obtain observation results, which can then be displayed.

[0027] Timely discover and respond to potential risks and challenges, and enhance the adaptability of systems or equipment to complex environments. The real-time data of sonar can instantly reflect the current underwater environmental conditions, quickly obtain observation results, and then continuously monitor changes in the underwater environment, timely discover and respond to potential risks and challenges, and enhance the adaptability of systems or equipment to complex environments.

[0028] In some embodiments, see Figure 1 The steps of constructing the preset fault prediction model are as follows: Acquire a historical data set, and process the historical data set to obtain a feature data set; Through the LSTM model, the mean square error is selected as the loss function, and the RMSprop optimizer is used to train with the feature data set to obtain a fault prediction model.

[0029] LSTM (Long Short-Term Memory) is a special type of recurrent neural network (RNN) that is particularly good at processing time series data with long-term dependencies. In fault warning, time series data (such as equipment operating status, sensor readings, etc.) is a common input data. LSTM can capture the time dependency and dynamic changes in these data, thereby more accurately predicting the future system status.

[0030] The mean square error is the mean of the sum of squares of the differences between the model's predicted values ​​and the actual values. It can well reflect the model's prediction accuracy. By penalizing large mistakes made by the model through squared error, the model can pay more attention to samples with larger prediction errors during training, thereby optimizing the prediction results.

[0031] The RMSprop optimizer can adaptively adjust the learning rate of each parameter according to the size of the gradient, thereby reducing oscillations and instability during training. During the training of the fault prediction model, this helps the model converge to the optimal solution faster. The learning rate is adjusted by calculating the exponentially weighted moving average of the squared gradient, so that the model performs well when dealing with non-stationary objective functions. Compared with traditional gradient descent algorithms, RMSprop can converge to the optimal solution faster and has better generalization ability. This helps to train a better performing fault prediction model in a limited time.

[0032] By combining the LSTM model, mean square error loss function and RMSprop optimizer, a fault prediction model with high accuracy, high stability and high training efficiency can be constructed. It can accurately predict the future state of the system, detect potential faults in time, and provide strong guarantee for the stable operation of the system.

[0033] In some embodiments, see Figure 1 , comparing the first prediction result with the real-time data to obtain the fault characteristics comprises: Comparing the first prediction result with the real-time data and calculating the residual; According to a preset residual distribution threshold, residuals exceeding the preset threshold are regarded as fault features.

[0034] By clarifying the connection obtained by comparing the first prediction result with the real-time data, it is possible to directly locate whether the root cause of the fault comes from the target object, thereby avoiding the possibility of blind investigation and misdiagnosis.

[0035] The residual is the difference between the actual observation and the model prediction. By calculating and analyzing the residual, we can evaluate how well the model fits the data. Abnormal residuals are usually a direct reflection of abnormal observations. When the residual of an observation deviates significantly from the residuals of other observations, the observation is abnormal. Using such abnormal observations as fault features for subsequent prediction can improve the quality of the final prediction results.

[0036] In some embodiments, see Figure 1 The step of performing cluster analysis on the fault features to obtain a second prediction result comprises: Cluster the fault features, classify similar fault features into the same category, and obtain clustering results; The clustering result is input into a prediction model based on clustering to perform prediction, thereby obtaining a second prediction result.

[0037] Cluster analysis can identify hidden patterns and structures in data, and cluster fault data with similar characteristics together to form different fault categories. This makes fault identification more accurate and reduces the possibility of false positives and false negatives. Based on the second prediction result obtained by cluster analysis, the original fault prediction model is improved and optimized to improve the prediction accuracy and generalization ability of the present invention. Among them, the cluster-based prediction model can be constructed by selecting any existing cluster analysis model.

[0038] In some embodiments, see Figure 1 The step of issuing a fault alarm based on the second prediction result includes: In response to the second prediction result being a fault caused by the target object, directly triggering an alarm; In response to the second prediction result being that the fault is not caused by the target object, the fault type is reported, and maintenance is performed based on the fault type.

[0039] The second prediction result indicates that the system status failure is caused by interference or collision with an underwater target, which directly issues an alarm so that the staff can consider taking measures such as avoidance, tracking or expulsion.

[0040] If it is not the impact of the target object, it means that the ship itself has a fault and the staff needs to be notified for maintenance.

[0041] The present invention proposes a sonar analysis and fault warning system with a domestically produced architecture. Figure 2 ,include: An acquisition unit 100 is configured to acquire real-time data and perform reasoning on the real-time data to obtain observation results; The prediction unit 200 is configured to input the real-time data and the observation result into a preset fault prediction model to obtain a first prediction result; The clustering unit 300 is configured to compare the first prediction result with the real-time data to obtain a fault feature, and perform cluster analysis on the fault feature to obtain a second prediction result; The alarm unit 400 is configured to issue a fault alarm based on the second prediction result.

[0042] On the one hand, the present invention can monitor underwater targets in real time, and on the other hand, it can predict faults when the system is about to become abnormal. Once an abnormality is found, it can give an early warning and automatically issue an alarm in real time to remind monitoring personnel to perform maintenance in time, thereby realizing real-time monitoring, target identification and fault prediction of the sonar system. It is efficient, stable and reliable, and can significantly improve the performance and safety of the sonar system.

[0043] See below Figure 3 and Figure 4 , a system architecture that can implement the concept of the present invention can be implemented as follows: Figure 3 As shown in the figure, the sonar intelligent analysis and fault prediction hardware system with a domestic architecture includes 1 computing module, 1 sonar signal acquisition module, 1 health management module, 2 domestic GPU modules, 1 storage module, 1 backplane module, and 1 power supply module.

[0044] Among them, the power supply module mainly powers the backplane module, computing module, sonar signal acquisition module, sonar system health management module, domestic GPU module, storage module, etc.

[0045] The computing module is connected to the sonar signal acquisition module, GPU module 1, GPU module 2, and storage module through the backplane module via the CPU's PCIE signal. The health management module is connected to the computing module via a gigabit network signal. The status information of each module in the system is monitored in real time via I2C signals, including voltage, current, temperature, etc. The storage module expands the domestic NVME hard disk through 8 M.2 interfaces, with a total storage capacity of 16TB, providing storage services for the entire system.

[0046] The computing module is designed with the domestic Tengyun S5000C processor (32 cores), with 64GB of surface-mounted memory, and a domestic graphics card (JH930) is expanded through PCIE. The sonar signal acquisition module is implemented with the domestic Fudan Micro FPGA as the main control design, and is connected to the computing module through PCIE signals. The multi-channel sonar signal is first conditioned by amplification and filtering, and then enters the FPGA after analog-to-digital conversion by the analog-to-digital conversion unit. The domestic GPU module is designed with the Zhikai 100 chip and is connected to the computing module through PCIE signals. The computing module sends the signal processed by the sonar acquisition module to the GPU module, and uses the domestic GPU module to perform pre-processing such as denoising, filtering, and FFT transformation on the signal data, extract key features, and analyze the signal data in combination with the sonar signal classification and recognition model that has been deployed in the early stage, so as to realize automatic recognition and classification of underwater targets. The storage module adopts 8 2TB M.2 NVME designs, and the computing module expands the storage module through PCIE signals, achieving a total storage capacity of 16TB. It is mainly used for the storage of signal data. The health management module uses a domestic embedded processor as the main control CPU, and monitors the CPU voltage, CPU temperature, memory voltage, memory temperature, graphics card voltage, and graphics card temperature of the computing module in real time through I2C; monitors the voltage, current, and temperature of the sonar signal acquisition module; monitors the voltage and temperature of the GPU module; monitors the voltage and temperature of the storage module; and monitors the voltage, current, and temperature of the power supply module. The module has an external gigabit network interface for user web access. The health management module and the computing module are interconnected through a gigabit network. The power supply module provides power supply requirements for the hardware system, and realizes power input to the computing module, sonar signal acquisition module, sonar system health management module, domestic GPU module, and storage module through the backplane module.

[0047] refer to Figure 4 The prediction mechanism of the present invention is a double-layer prediction mechanism, which is based on the monitoring history data collected by the health management module in the sonar system, that is, the previous monitoring data of the system (including normal and abnormal data), the historical sonar data collected by the sonar signal acquisition module, and the sonar analysis and recognition results corresponding to the calculation module to form a data set A.

[0048] After preprocessing the data set A (including denoising, abnormal data cleaning, normalization, etc.), the fault-related features (voltage, current, temperature abnormal values, and abnormal values ​​of sonar data and sonar recognition results) are extracted, and the processed data set A is feature constructed based on sliding window statistics and other methods to obtain data set B. The long short-term memory network LSTM is selected as the basic model, the mean square error method is used as the loss function, and RMSprop is used as the optimizer. Training is performed based on data set B to obtain the fault prediction model A of the system.

[0049] The data collected in real time by the health management module and the target object data identified by the calculation module are sent to the fault prediction model A after preprocessing to obtain the prediction result C.

[0050] The target results identified by sonar and health monitoring data are used as actual observation data, and the prediction result C is obtained after historical data prediction. The prediction result C is compared with the actual observation data to determine whether the target identified by sonar will affect the system state. In order to further analyze the relationship between the impact and the prediction result C, residuals or outliers are extracted as fault features. The residuals are calculated to compare the actual observation data and the prediction result C point by point, and the difference between the two is calculated to reflect the degree of difference. The outliers corresponding to the extracted residuals are standardized to eliminate the influence of different dimensions and magnitudes on data analysis, and cluster analysis is performed. In addition to considering the faults caused by the internal system state, this processing also increases the consideration of external targets. External targets are used as a category of cluster analysis, which effectively expands the scope of monitoring in practical applications, more comprehensively optimizes the early warning function of the system, and facilitates targeted maintenance when a fault occurs.

[0051] Select appropriate clustering algorithms, such as K-means, hierarchical clustering, DBSCAN, etc., use clustering analysis technology to cluster fault features, use silhouette coefficient, elbow rule and other methods to determine the optimal number of clusters, classify similar fault modes into the same category, analyze the clustering results, identify different fault features, and analyze whether the clustering results meet expectations and whether they can accurately reflect the fault features in the data.

[0052] The clustering results are explained to clarify the fault characteristics represented by each cluster. The result D obtained after the cluster analysis model A is judged. When the prediction result D indicates that a fault is about to occur, the possible fault type is predicted based on these fault characteristics, and the alarm rules are triggered to notify maintenance personnel to come for maintenance.

[0053] When performing sonar analysis, the target objects monitored by the sonar can also be displayed to ensure that the staff can observe the target objects in front. The specific sonar analysis process is as follows: During the process of ocean detection, underwater target identification, marine environment monitoring, etc., the ship uses the sonar signal acquisition module to amplify and filter the collected multi-channel sonar signals, and then performs signal conditioning such as analog-to-digital conversion through its internal analog-to-digital conversion unit and sends it to the calculation module. The calculation module sends the signal data to the GPU module 1, and extracts the key features after pre-processing such as denoising, filtering, and FFT transformation. At the same time, the original signal data is stored in the storage module. The calculation module calls the GPU module 1 to run the sonar signal classification and recognition model inference engine server deployed in the early stage, performs real-time intelligent analysis of the key features of the signal, realizes automatic identification and classification of underwater targets (fish, submersibles, etc.), and relies on the graphics card in the calculation module for real-time display for real-time viewing by personnel.

[0054] During sonar analysis, sonar data is also used as part of the fault warning monitoring data, managing the internal and external influences of the system at the same time, effectively avoiding the impact of single monitoring on the difficulty of fault detection, and timely reminding the staff to take targeted measures. The fault warning process is as follows: During the process of ocean detection, underwater target identification, marine environment monitoring, etc., the health management module will collect the voltage, current, temperature and other status information of each module (CPU voltage, CPU temperature, memory voltage, memory temperature, graphics card voltage, graphics card temperature of the computing module; voltage, current, temperature of the sonar signal acquisition module; voltage and temperature of the GPU module; voltage and temperature of the storage module; voltage, current, temperature of the power supply module) and send it to the computing module. The computing module calls GPU module 2 to run the fault prediction model A in real time, and the real-time status information obtained is preprocessed and passed through the fault prediction model A to obtain the prediction result C. The prediction result C and the cluster analysis model A that are run in real time are combined to obtain the prediction result D. When the prediction result D indicates that a fault is about to occur, the possible fault type is reported and the alarm rule is triggered. After the alarm is driven by the alarm system to sound, the onboard monitoring personnel and other personnel will quickly take specific maintenance measures according to the situation.

[0055] Through the above methods, the ship can maintain its running state whether it is in operation or docking. On the one hand, it can monitor underwater targets in real time, and on the other hand, it can predict the failure of the system when it is about to be abnormal. Once it is found, it will be warned in advance and automatically send out an alarm in real time to remind the monitoring personnel to carry out maintenance in time.

[0056] As an optimization, the Gigabit network used to communicate between the health management module and the computing module can be replaced with a higher-speed 40G or 100G Ethernet network.

[0057] Based on the same inventive concept, according to another aspect of the present invention, Figure 5 As shown, an embodiment of the present invention further provides a computer device 30, which includes a processor 310 and a memory 320. The memory 320 stores a computer program 321 that can be run on the processor. When the processor 310 executes the program, the steps of the above method are performed.

[0058] Based on the same inventive concept, according to another aspect of the present invention, Figure 6 As shown, an embodiment of the present invention further provides a computer-readable storage medium 40, which stores a computer program 410 for executing the above method when executed by a processor.

[0059] The embodiment of the present invention may also include a corresponding computer device. The computer device includes a memory, at least one processor, and a computer program stored in the memory and executable on the processor, and the processor executes any one of the above methods when executing the program.

[0060] The memory, as a non-volatile computer-readable storage medium, can be used to store non-volatile software programs, non-volatile computer executable programs and modules, such as program instructions / modules in the embodiments of the present application. The processor executes various functional applications and data processing of the device by running the non-volatile software programs, instructions and modules stored in the memory, that is, implementing the above method.

[0061] The memory may include a program storage area and a data storage area, wherein the program storage area may store an operating system, an application required for at least one function; the data storage area may store data created according to the use of the device, etc. In addition, the memory may include a high-speed random access memory, and may also include a non-volatile memory, such as at least one disk storage device, a flash memory device, or other non-volatile solid-state storage device. In an embodiment, the memory may optionally include a memory remotely arranged relative to the processor, and these remote memories may be connected to the local module via a network. Examples of the above-mentioned network include, but are not limited to, the Internet, an intranet, a local area network, a mobile communication network, and combinations thereof.

[0062] Finally, it should be noted that a person skilled in the art can understand that all or part of the processes in the above-mentioned embodiments can be implemented by instructing the relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium. When the program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, the storage medium of the program can be a disk, an optical disk, a read-only storage memory (ROM) or a random access memory (RAM), etc. The above-mentioned computer program embodiments can achieve the same or similar effects as the corresponding above-mentioned arbitrary method embodiments.

[0063] It will also be appreciated by those skilled in the art that various exemplary logic blocks, modules, circuits and algorithm steps described in conjunction with the disclosure herein can be implemented as electronic hardware, computer software or a combination of the two. In order to clearly illustrate this interchangeability of hardware and software, a general description has been given to the functions of various schematic components, blocks, modules, circuits and steps. Whether this function is implemented as software or hardware depends on specific applications and the design constraints imposed on the entire system. Those skilled in the art can implement the function in various ways for each specific application, but this implementation decision should not be interpreted as causing a departure from the disclosed scope of the embodiments of the present invention.

[0064] The above are exemplary embodiments disclosed in the present invention, but it should be noted that various changes and modifications may be made without departing from the scope of the embodiments disclosed in the present invention as defined in the claims. The functions, steps and / or actions of the method claims according to the disclosed embodiments described herein do not need to be performed in any particular order. The serial numbers of the embodiments disclosed in the above embodiments of the present invention are for description only and do not represent the advantages and disadvantages of the embodiments. In addition, although the elements disclosed in the embodiments of the present invention may be described or required in individual form, they may also be understood as multiple unless explicitly limited to the singular.

[0065] It should be understood that, as used herein, the singular forms "a", "an" are intended to include the plural forms as well, unless the context clearly supports an exception. It should also be understood that, as used herein, "and / or" refers to any and all possible combinations including one or more of the associated listed items.

[0066] A person skilled in the art should understand that the discussion of any of the above embodiments is only exemplary and is not intended to imply that the scope of the disclosure of the embodiments of the present invention (including the claims) is limited to these examples; under the concept of the embodiments of the present invention, the technical features in the above embodiments or different embodiments can also be combined, and there are many other changes in different aspects of the above embodiments of the present invention, which are not provided in detail for the sake of simplicity. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the embodiments of the present invention should be included in the protection scope of the embodiments of the present invention.

Claims

1. A method for sonar analysis and fault warning of a domestically produced structure, characterized in that: include: Acquire real-time data, and perform reasoning on the real-time data to obtain observation results; Inputting the real-time data and the observation result into a preset fault prediction model to obtain a first prediction result; Comparing the first prediction result with the real-time data to obtain fault characteristics, and performing cluster analysis on the fault characteristics to obtain a second prediction result; A fault alarm is issued based on the second prediction result.

2. The method for sonar analysis and fault warning of a domestically produced structure according to claim 1, characterized in that: The real-time data includes one or more of multi-channel sonar signals and system status data; wherein the system status data includes one or more of voltage, current and temperature.

3. The method for sonar analysis and fault warning of a domestically produced structure according to claim 2, characterized in that: The step of inferring the real-time data to obtain the observation result comprises: Performing signal conditioning and analog-to-digital conversion on the multi-channel sonar signals in the real-time data; The converted signal is stored and preprocessed to obtain key features; Real-time reasoning is performed on the key features to obtain observation results.

4. The method for sonar analysis and fault warning of a domestically produced structure according to claim 2, characterized in that: The steps of constructing the preset fault prediction model are as follows: Acquire a historical data set, and process the historical data set to obtain a feature data set; Through the LSTM model, the mean square error is selected as the loss function, and the RMSprop optimizer is used to train with the feature data set to obtain a fault prediction model.

5. The method for sonar analysis and fault warning of all domestically produced structures according to claim 1 is characterized in that: The step of comparing the first prediction result with the real-time data to obtain the fault characteristics includes: Comparing the first prediction result with the real-time data and calculating the residual; According to a preset residual distribution threshold, residuals exceeding the preset threshold are regarded as fault features.

6. The method for sonar analysis and fault warning of all domestically produced structures according to claim 1, characterized in that: The step of performing cluster analysis on the fault features to obtain a second prediction result comprises: Cluster the fault features, classify similar fault features into the same category, and obtain clustering results; The clustering result is input into a prediction model based on clustering to perform prediction, thereby obtaining a second prediction result.

7. The method for sonar analysis and fault warning of a domestically produced structure according to claim 1, characterized in that: The step of issuing a fault alarm based on the second prediction result comprises: In response to the second prediction result being a fault caused by the target object, directly triggering an alarm; In response to the second prediction result being that the fault is not caused by the target object, the fault type is reported, and maintenance is performed based on the fault type.

8. A domestically-built sonar analysis and fault warning system, characterized by: include: An acquisition unit, configured to acquire real-time data and perform reasoning on the real-time data to obtain an observation result; A prediction unit, configured to input the real-time data and the observation result into a preset fault prediction model to obtain a first prediction result; A clustering unit, configured to compare the first prediction result with the real-time data to obtain a fault feature, and perform cluster analysis on the fault feature to obtain a second prediction result; An alarm unit is configured to issue a fault alarm based on the second prediction result.

9. A computer device comprising: at least one processor; and a memory storing a computer program executable on the processor, wherein the processor executes the steps of the method for sonar analysis and fault warning of a domestically produced architecture as described in any one of claims 1 to 7 when executing the program.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method for sonar analysis and fault warning of a domestically produced architecture as described in any one of claims 1 to 7 are performed.

Images