Ship equipment management system and method based on artificial intelligence

Abstract

The invention discloses a ship equipment management system and method based on artificial intelligence, and the system comprises a data collection module which collects the operation data of all equipment on a ship in real time through multiple types of sensors; the data storage module is used for storing and managing the acquired equipment data; the data analysis module is used for analyzing and predicting the equipment data by using a machine learning algorithm; the maintenance suggestion module is used for providing specific maintenance operation suggestions and maintenance plans according to the analysis result; the user interface module is used for providing a visual operation interface and displaying the operation state, historical operation data, maintenance records and maintenance operation guidance of the equipment in real time; and the security management module is used for encrypting the stored and transmitted data and setting an access permission system. According to the invention, by integrating the Internet of Things technology, the machine learning algorithm and the user-friendly operation interface, intelligent management of equipment is realized.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of equipment management, and particularly relates to a ship equipment management system and method based on artificial intelligence. BACKGROUND

[0002] With the rapid development of the shipping industry, the scale and complexity of ships are increasing, and the types and quantities of equipment on board are also increasing. Ship equipment refers to various devices and tools used for shipbuilding, operation and maintenance. These devices cover ship outfitting, navigation, power, safety, communication, repair and other aspects. How to efficiently manage these equipment and ensure their normal operation has become an important challenge in ship operation. The traditional equipment management mode relies on manual inspection and experience judgment, which has low efficiency and insufficient accuracy.

[0003] In recent years, the rapid development of artificial intelligence technology has provided a new solution for equipment management. However, the existing equipment management system still has the following shortcomings in terms of intelligence and predictive maintenance: (1) Limited data collection and processing capabilities, difficult to meet the needs of large-scale equipment management; (2) Lack of intelligent analysis and prediction capabilities, unable to identify potential failures in advance; (3) User interface is not user-friendly, operation is complex, and it is difficult to meet the actual needs of the crew; (4) System security is insufficient, vulnerable to data leakage and system failure.

[0004] Therefore, it is urgent to design a ship equipment management system and method based on artificial intelligence to solve the above problems of the prior art. SUMMARY

[0005] In view of the above defects or improvement needs of the prior art, the present application provides a ship equipment management system and method based on artificial intelligence, which can monitor the equipment status in real time, predict potential failures, provide maintenance recommendations, and optimize maintenance plans, thereby significantly improving the efficiency and accuracy of equipment management.

[0006] In order to achieve the above purpose, the present application adopts the following technical solutions: The present application provides a ship equipment management system based on artificial intelligence, comprising: a data acquisition module for real-time acquisition of operation data of each equipment on board through multiple types of sensors; a data storage module for storing and managing equipment data collected by the data acquisition module; a data analysis module for analyzing and predicting equipment data using machine learning algorithms; a maintenance suggestion module for providing specific maintenance operation suggestions and maintenance plans according to the analysis results of the data analysis module; a user interface module for providing an intuitive operation interface, displaying the running state of the equipment in real time, and providing historical running data, maintenance records and maintenance operation guidance of the equipment; a security management module for ensuring the security of system data and stable operation of the system by encrypting the stored and transmitted data and setting an access permission system.

[0007] As an embodiment of the present application, the data collection module comprises: a multi-type sensor system for monitoring the running state of the equipment by installing multiple types of sensors in combination with the classification of the ship equipment; a data collection unit for aggregating and pre-processing the equipment data collected by the multi-type sensor system through edge computing; a data transmission unit for transmitting the equipment data to the data storage module through wireless communication technology.

[0008] As an embodiment of the present application, the multi-type sensor system comprises: a vibration sensor for monitoring the vibration amplitude and frequency of the equipment; a temperature sensor for monitoring the temperature change of the equipment; a pressure sensor for monitoring the pressure change of the equipment; a humidity sensor for monitoring the change of environmental humidity; a displacement sensor for monitoring the change of the gap between components; a strain sensor for monitoring the structural stress; a rotational speed sensor for monitoring the rotational speed of the motor; a current sensor for monitoring the load current fluctuation; a flow sensor for monitoring the medium flow of the pipeline.

[0009] As an embodiment of the present application, the data storage module comprises: a structured database for storing the basic information and maintenance records of the equipment, wherein the basic information of the equipment includes the equipment model, manufacturer, installation location and technical parameters; a distributed storage unit for storing the running data of the equipment, realizing horizontal expansion and high availability through a distributed architecture.

[0010] As an embodiment of the present application, the data analysis module comprises: a data preprocessing module for cleaning and preprocessing the equipment data preprocessed through edge computing to remove noise data and outliers; Feature extraction module: used to extract relevant features from the equipment data processed by the data preprocessing module, including equipment running time, vibration frequency, temperature change; Machine learning modeling: different machine learning algorithms are used to model the equipment data after feature extraction for different task types, to predict the health status and failure risk of the equipment; Abnormality detection module: set threshold and abnormality detection algorithm to identify abnormal conditions in equipment operation and issue timely alerts.

[0011] As an embodiment of the present application, the different task types include fault classification, multi-classification, life prediction, image recognition, and the corresponding machine learning algorithms include support vector machine, random forest, long-term segment memory network, and convolutional neural network.

[0012] As an embodiment of the present application, the threshold setting includes: Static threshold: set absolute threshold based on equipment safety specifications; Dynamic threshold: generate adaptive threshold through historical data statistical analysis; Three-level early warning system: yellow warning: parameters exceed normal range by 10-20%, prompt attention; orange warning: parameters exceed normal range by 20-50%, arrange for inspection; red alert: parameters exceed normal range by more than 50% or trigger safety interlock condition, immediately shut down for maintenance.

[0013] As an embodiment of the present application, the encryption processing includes: Transport layer encryption: use TLS1.3 protocol to encrypt data transmission between sensors and edge nodes, edge nodes and cloud, key exchange uses ECDHE algorithm, data encryption uses AES-256-GCM; Storage layer encryption: use AES-256-CBC mode to encrypt sensitive fields, use homomorphic encryption technology to encrypt database backup files.

[0014] As an embodiment of the present application, the access permission system is based on the RBAC model and authorizes different data access permissions and operation permissions according to different roles.

[0015] The present application also provides a ship equipment management method based on artificial intelligence, which uses the above-mentioned ship equipment management system based on artificial intelligence, and specifically includes the following steps: S1: Real-time acquisition of running data of each equipment on the ship through multiple types of sensors; S2: Store and manage the collected equipment data; S3: Analyze and predict the equipment data using machine learning algorithms; S4: providing specific maintenance operation suggestions and maintenance plans according to the analysis results; S5: displaying the running state, historical running data, maintenance records and maintenance operation guidance of the equipment in real time through a user interface; S6: encrypting the stored and transmitted data and setting an access permission system.

[0016] The beneficial effects of the present application are: 1. The present application collects the running data of each equipment on the ship through multiple types of sensors, stores the data using a structured database and a distributed storage unit, models the equipment data using a machine learning algorithm, predicts the health status and failure risk of the equipment, and solves the problems of efficiency, accuracy and safety in traditional ship equipment management through a user-friendly interface and a safety mechanism, realizing intelligent management of the equipment, improving management efficiency and accuracy, and solving the problem of limited data collection and processing capacity in existing equipment management systems, which cannot meet the demand of large-scale equipment management; 2. The present application preprocesses the equipment data through a data analysis module, extracts features, models the equipment data using different machine learning algorithms for different task types, predicts the health status and failure risk of the equipment, can identify potential failures in advance, reduces equipment downtime and maintenance costs, and solves the problem of lack of intelligent analysis and prediction capability in existing equipment management systems, which cannot identify potential failures in advance; 3. The present application provides an intuitive operation interface, displays the running state of the equipment in real time, including the health status of the equipment, failure alarms, provides historical running data and maintenance records of the equipment, facilitates the crew to check the usage and maintenance history of the equipment, provides detailed steps and guidance for maintenance operations, helps the crew to quickly complete maintenance tasks, improves user experience, and solves the problem of existing equipment management systems that have an unfriendly user interface, complex operation, and cannot meet the actual needs of the crew; 4. The present application encrypts the stored and transmitted data through a safety management module to prevent data from being illegally obtained and tampered with, and sets a strict access permission system to ensure that only authorized personnel can access system data and perform critical operations, monitors the running state of the system in real time, discovers and handles system failures and abnormal conditions in a timely manner, ensures the security of system data and the stable operation of the system, prevents data leakage and system failure, and solves the problem of existing equipment management systems that have insufficient security and are vulnerable to data leakage and system failure. BRIEF DESCRIPTION OF DRAWINGS

[0017] Figure 1 The overall architecture diagram of the ship equipment management system based on artificial intelligence provided in the embodiments of the present application is shown in the figure; Figure 2A data acquisition module flowchart of an artificial intelligence-based ship equipment management system provided in an embodiment of the present application; Figure 3 A data analysis module flowchart of an artificial intelligence-based ship equipment management system provided in an embodiment of the present application; Figure 4 A user interface module schematic diagram of an artificial intelligence-based ship equipment management system provided in an embodiment of the present application. DETAILED DESCRIPTION

[0018] The technical solutions in the embodiments of the present application will be described clearly and completely below with reference to the drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0019] Referring to Figures 1-4 The present application provides an artificial intelligence-based ship equipment management system, comprising: A data acquisition module for real-time acquisition of operation data of each equipment on the ship through multiple types of sensors; A data storage module for storing and managing equipment data collected by the data acquisition module; A data analysis module for analyzing and predicting equipment data using machine learning algorithms; A maintenance suggestion module for providing specific maintenance operation suggestions and maintenance plans according to the analysis results of the data analysis module; A user interface module for providing an intuitive operation interface, real-time display of equipment operation status, and provision of historical operation data, maintenance records and maintenance operation guidance of equipment; A safety management module for ensuring the security of system data and stable operation of the system by encrypting the stored and transmitted data and setting an access permission system.

[0020] The data acquisition module in the present application covers mechanical, electrical and environmental parameters through targeted deployment of multiple types of sensors, and real-time acquisition of operation data of each equipment on the ship, including equipment status, usage, environmental parameters, etc. Multiple types of sensors are installed at key parts of the equipment, such as temperature sensors, vibration sensors, pressure sensors, etc. The data acquisition module aggregates and performs edge computing preprocessing on sensor data, and transmits the data to a central server or cloud storage through wireless communication technologies such as Wi-Fi, Bluetooth, 4G / 5G.

[0021] Among them, the collected equipment status includes mechanical status and running status, and the specific corresponding sensor types are: Mechanical state: adopt vibration sensor, such as three-axis acceleration sensor, to monitor equipment vibration amplitude, frequency, identify bearing wear, gear failure, etc.; displacement sensor, such as laser range finder, to monitor component gap change, such as valve spool displacement; strain sensor to monitor structural stress, such as stress concentration at key connection of ship body; Running state: adopt speed sensor, such as electromagnetic or photoelectric, to monitor motor / pump speed; temperature sensor, such as PT100 platinum resistance, to monitor bearing, motor winding, pipeline temperature; pressure sensor, such as piezoresistive transducer, to monitor hydraulic system pressure, cylinder compression pressure.

[0022] The sensor type corresponding to the use case is: Load parameter: adopt current / voltage sensor to monitor equipment power consumption, reflecting load rate; adopt ultrasonic or vortex flow sensor to monitor pipeline medium flow, such as fuel, cooling water; adopt speed sensor or current sensor to accumulate running time.

[0023] The sensor type corresponding to the environmental parameter is: Environmental state: adopt temperature and humidity sensor to monitor cabin, equipment cabin environmental temperature and humidity, prevent electrical equipment from being damp; barometric pressure sensor to monitor atmospheric pressure, assist ship stability calculation; harmful gas sensor, such as CO, Sensor, to ensure operation safety.

[0024] As an embodiment of the present application, the data acquisition module comprises: Multi-type sensor system: install multi-type sensors in combination with ship equipment classification to monitor the running state of equipment; the multi-type sensor installation in combination with ship equipment classification is specifically: Power equipment (such as main engine, generator): Vibration sensor: installed on bearing seat, gear box shell (axial / radial / tangential three directions); Temperature sensor: embedded in motor winding, bearing grease cavity, cooling water pipeline surface; Pressure sensor: installed on cylinder head to monitor burst pressure; lubricating oil main oil way to monitor oil supply pressure.

[0025] Outfitting equipment (such as valve, pump): Displacement sensor: installed on valve actuator to monitor valve core opening; Flow sensor: installed on pump inlet and outlet pipeline, preferably horizontal straight pipe section, to ensure measurement accuracy; Pressure sensor: installed on pump outlet valve after pipeline to monitor output pressure.

[0026] Navigation and safety equipment (e.g. radar, life-saving equipment): State sensor: installed on the radar antenna shaft, used to monitor the rotation angle / rate; life-saving valve sealing surface temperature / humidity sensor, used to monitor the sealing performance; Environmental sensor: installed on the external shell of the equipment, used to monitor the temperature and humidity of the open environment, salt spray corrosion parameters.

[0027] Electrical equipment (e.g. cables, power distribution cabinets): Temperature sensor: wrapped around the cable joint, used to monitor the temperature rise caused by abnormal contact resistance; Current sensor: connected in series with the main circuit, used to monitor the load current fluctuation.

[0028] Data acquisition unit: collects and pre-processes the equipment data collected by the multi-type sensor system, obtains structured time series data, feature preliminary extraction data and compressed and packaged data, specifically: Structured time series data: contains time stamps accurate to milliseconds, ensures the time sequence consistency of multi-sensor data; sensor ID, equipment number, establishes a three-level index of equipment-sensor-parameter, facilitates quick fault source positioning, such as retrieving the full life cycle data of a certain pump through the equipment number; measurement parameter value, such as vibration acceleration value, temperature value; unit, such as g, ℃; data quality identifier, marks data validity, such as normal / abnormal / missing, avoids invalid data pollution of analysis model, such as automatically skipping the time period of sensors marked as "abnormal"; Feature preliminary extraction data: filtering and noise reduction through edge computing equipment, such as Butterworth low-pass filter to remove high-frequency noise and retain valid signals; outlier marking, identifies outliers based on IQR algorithm or 3σ principle; data normalization, unifies dimensions for subsequent analysis; Compressed and packaged data: lossy compression of large data volume parameters such as high-frequency vibration signals, such as wavelet compression, reduces data volume by 50%-70% under the premise of ensuring relevant features are not lost, reduces transmission load.

[0029] Among them, edge computing preprocessing is a filtering gateway from sensor to system core processing link, mainly used to improve data quality, unify data format, reduce transmission pressure and reduce cloud load, etc., to solve the availability, transmission efficiency and computing load problems of raw data, the structured time series data is used to build a data space-time coordinate system, realize full-process tracing and accurate analysis, the feature preliminary extraction data extracts physical semantics from raw signals, reduces data dimension and highlights key information, the compressed and packaged data balances data integrity and transmission efficiency, adapts to complex communication environment of ships.

[0030] Data transmission unit: transmit equipment data to data storage module through wireless communication technology.

[0031] Specifically, the multi-type sensor system, data acquisition unit and data transmission unit are seamlessly connected through diversified physical interfaces, hierarchical network protocols of industrial bus + TCP / IP + cloud protocol and intelligent routing strategies of near field priority + far field backup, forming a complete link from sensor data acquisition to cloud storage.

[0032] As an embodiment of the present application, the multi-type sensor system comprises: Vibration sensor: used for monitoring equipment vibration amplitude, frequency; Temperature sensor: used for monitoring equipment temperature change; Pressure sensor: used for monitoring equipment pressure change; Humidity sensor: used for monitoring environmental humidity change; Displacement sensor: used for monitoring component gap change; Strain sensor: used for monitoring structure stress; Rotational speed sensor: used for monitoring motor rotational speed; Current sensor: used for monitoring load current fluctuation; Flow sensor: used for monitoring pipeline medium flow.

[0033] The data storage module of the present application stores basic information, running data and maintenance records of equipment by designing a structured database; adopts distributed storage technology (such as Hadoop, cloud storage) to store a large amount of equipment data, ensuring data security and scalability, regularly backing up data to prevent data loss and ensure stable operation of the system.

[0034] As an embodiment of the present application, the data storage module comprises: Structured database (such as MySQL, PostgreSQL): mainly used for storing equipment basic information and maintenance records, the equipment basic information includes equipment model, manufacturer, installation location, technical parameters; the maintenance records include repair time, replaced parts, repair personnel, etc.; these data closely fit the ship industry standards, regulations and actual business processes, and customized solutions are designed on demand, with clear table structure and relationship model, suitable for high-frequency add, delete, modify and query operations; Distributed storage unit (such as Hadoop HDFS, cloud storage): mainly used for storing massive equipment operation data, such as vibration, temperature, pressure and other time series data collected by sensors in real time. Such data is large in quantity and various in format, containing structured numerical values and unstructured waveform files, and needs to be expanded horizontally and highly available through a distributed architecture, which is a technology system that realizes data storage and processing through the cooperative work of multiple physical servers or cloud nodes. The core goal is to solve the problems of limited storage capacity, insufficient computing power and single point failure of a single machine. In this application, the architecture design of the distributed storage unit meets the needs of massive data storage, high concurrency access and low latency response.

[0035] The data management mode of the data storage module in this application includes structured data management, including equipment basic information and maintenance records; unstructured / massive data management, including sensor operation data; data life cycle management and data security management, specifically: Structured data management: A relational database such as MySQL or PostgreSQL is used to establish standardized data tables, including equipment archives table, maintenance record table and spare parts inventory table. Efficient queries can be achieved through SQL language, such as retrieving historical maintenance records by "equipment model + maintenance time", with a response time of <0.5 seconds. Support for foreign key association, such as the association between the equipment archives table and the maintenance record table through "device ID"; ensure data consistency, such as automatically cascading deletion of related maintenance records when deleting equipment.

[0036] Unstructured / massive data management: Distributed file system Hadoop HDFS or cloud storage, such as Aliyun OSS, is used to store sensor raw data, such as vibration waveform files and log data, in blocks according to the naming rule of timestamp + device ID, with a single block size of 128MB, supporting horizontal expansion, and a single cluster can carry 100,000+ device data. Through a metadata management system such as Hive Metastore, record data location, format, collection time and other information to achieve fast retrieval, such as locating a device's temperature data file from last week in seconds.

[0037] Data life cycle management: Cold and hot data classification: real-time data (up to 30 days) is stored in high-performance SSD servers, supporting millisecond-level real-time access; historical data (more than 30 days) is migrated to low-cost HDD or the cloud, and is managed uniformly through a data lake architecture such as AWS LakeFormation; Automatic backup strategy: structured database daily full backup + incremental backup (RPO≤15 minutes), distributed storage data fault tolerance is realized through multiple copy mechanism (default 3 copies), regular recovery exercise, 2 times disaster recovery test per year.

[0038] Data security management: Access control: limit IP access through database firewall such as MariaDB Firewall, enable bucket strategy such as Bucket Policy for cloud storage to control read and write permissions; Audit log: record all data operations (query, modify, delete) of users, time, IP address, store in independent audit database, retention period 5 years.

[0039] The data storage module in the application adopts the complementary fusion mode of structured database and distributed storage unit, and the structured database can be deployed on a distributed cluster, such as a distributed relational database TiDB, and the distributed technology is used to improve the storage capacity and query performance; The distributed storage system can support structured query through middleware Hive, realize the analysis and statistics of running data, such as aggregating the average temperature of equipment according to the time dimension, provide complete data source for subsequent machine learning model, and support complex analysis task.

[0040] As an embodiment of the application, the data analysis module comprises: Data preprocessing module: clean and preprocess the equipment data preprocessed by edge computing, remove noise data and outliers, specifically including: Noise removal: for high-frequency signals such as vibration and current, a Butterworth low-pass filter is used, and the cutoff frequency is set according to the inherent frequency of the equipment, such as the bearing fault characteristic frequency is usually <500Hz, filter the high-frequency noise outside the mechanical resonance, eliminate the measurement error of the sensor, avoid the noise covering the real fault characteristics; for slowly varying signals such as temperature and pressure, moving average filter is used, window size 5-10 minutes, smooth random fluctuation, eliminate sensor instantaneous jump, improve the accuracy of subsequent feature extraction, such as the FFT spectrum of the filtered vibration signal is clearer, and the fault frequency recognition accuracy is improved by 20%; Outlier detection: Static threshold method: set a safety range based on the factory parameters of the equipment, such as marking the bearing temperature >120℃ as abnormal; Statistical method: IQR (interquartile range) method (remove data points exceeding Q3+1.5IQR or Q1-1.5IQR), 3σ principle (remove data exceeding mean±3 times standard deviation); Machine learning method: Isolation Forest detects non-stationary anomalies (such as sudden vibration spikes) with an accuracy of 98%; Combining the above three anomaly detection methods, different types of anomalies can be covered. The static threshold method is suitable for parameters with known safety boundaries, such as temperature > 120℃ and response speed < 1ms. The statistical method uses adaptive dynamic data distribution, such as the vibration baseline change of the ship under different working conditions, with a false positive rate of < 5%. The machine learning method is used to capture non-linear anomaly patterns, such as intermittent vibration spikes, with a detection rate of early faults 15% higher than traditional methods.

[0041] Anomaly repair: linear interpolation is used to fill short-time missing data, such as < 10 minutes of temperature data missing. Long-time abnormal data, such as > 1 hour of vibration signal distortion, is marked as invalid to avoid introducing false features by interpolation and trigger sensor failure alarm.

[0042] Feature extraction module: used to extract relevant features from the equipment data processed by the data preprocessing module. It mainly compresses the original high-dimensional data into low-dimensional, physically meaningful features, with a 90% improvement in computational efficiency. It highlights features related to faults, such as bearing fault characteristic frequencies, and suppresses irrelevant information, such as environmental noise, to improve the accuracy of subsequent machine learning models by 30%. The relevant features extracted, such as temperature change rate, have physical meaning and are easy for engineers to understand the equipment state, such as temperature change rate > 0.5℃ / min indicating lubrication system anomaly. The relevant features include equipment running time, vibration frequency, temperature change, etc., specifically: Time domain features: calculate equipment running time by start-stop signal integration, calculate load rate by current-to-rated current ratio, and calculate peak factor by vibration peak-to-rms ratio to characterize equipment running intensity; Frequency domain features: FFT transform of vibration signal to extract bearing fault characteristic frequencies, such as rolling element pass frequency BPFI = 0.42 × speed × number of balls; gear meshing frequency, such as 2000rpm gear tooth number 30 corresponding to 1000Hz, to identify specific component faults; Derived features: combine environmental parameters, such as temperature increase of 10℃, equipment life reduction of 15%, generate life correction factor to improve prediction model accuracy; This application extracts relevant features using vibration signals as an example: Time domain features: mean, variance (reflecting vibration energy); kurtosis (impact indicator, kurtosis value > 3 when bearing fails); peak factor (wear failure > 5); Frequency domain features: extract the main frequency and multiple frequency components through FFT transformation, such as the 2-3 times frequency energy of the meshing frequency of gear failure; envelope spectrum analysis, demodulation of fault characteristic frequency, such as the outer ring fault frequency of rolling bearing; Time-frequency features: wavelet transform, identify time-varying frequency components, such as high-frequency noise mutation in the early stage of valve leakage; short-time Fourier transform (STFT) generates time-frequency matrix as input of deep learning model.

[0043] Machine learning modeling: different machine learning algorithms are used for different task types to model the equipment data after feature extraction, to predict the health status and failure risk of the equipment; Through the three-level processing flow of data preprocessing module, feature extraction module and machine learning modeling, the data input into the model has high reliability, strong representation and low redundancy, which significantly improves the accuracy and efficiency of intelligent analysis.

[0044] As an embodiment of the present application, the different task types include fault classification, multi-classification, life prediction, image recognition, and the corresponding machine learning algorithms include support vector machine, random forest, long short-term memory network and convolutional neural network.

[0045] Among them, the algorithm selection basis is shown in Table 1:

[0046] Table 1 Algorithm selection basis Fault classification task algorithm selection basis: Support vector machine (SVM) is used, radial basis kernel function (RBF) is used to handle nonlinear boundary, and grid search is used to optimize penalty parameter C (range 1-100) and kernel function parameter γ (range 0.1-10); Multi-classification task algorithm selection basis: It can handle high-dimensional features, such as 200+ dimensional features of vibration signals, reduce the risk of overfitting through ensemble learning such as voting of multiple decision trees, and automatically filter key features such as Gini index, such as the contribution of kurtosis and peak factor to wear classification is more than 80%; It is robust to unbalanced data, such as crack fault samples accounting for 5%, and balances the classes through undersampling / oversampling techniques such as SMOTE algorithm.

[0047] Life prediction task algorithm selection basis: Use long short-term memory network (LSTM), the number of input layer neurons is equal to the feature dimension (such as 10 sensor parameters), the hidden layer has 2 layers with 128 units each, and the output layer predicts the health value (0-100 points) in the next 24 hours.

[0048] Image recognition task algorithm selection basis: Convolutional layers can automatically extract local features of time-frequency diagrams, such as high-frequency energy concentration areas corresponding to bearing failures, reducing the cost of manual feature engineering; pooling layers achieve feature dimensionality reduction and translation invariance, such as fault features at different positions can be identified, improving model generalization ability; suitable for processing two-dimensional data such as STFT generated time-frequency matrix, with a 90% higher computational efficiency than fully connected networks.

[0049] The model training process is as follows: Data division: divide the equipment data into training set, validation set and test set according to the ratio of 7:2:1, ensure sample balance (fault sample ratio ≥20%), training set for model fitting, validation set for adjusting hyperparameters, test set for evaluating generalization ability; Optimization goal: cross-entropy loss function for classification task, mean square error (MSE) for regression task, using Adam optimizer (learning rate 0.001) for 500 iterations; Model evaluation: classification model needs to achieve accuracy ≥95%, recall rate ≥90%, regression model needs to meet R²≥0.85 to be deployed.

[0050] Abnormality detection module: set threshold and abnormality detection algorithm to identify abnormal conditions in equipment operation and issue alerts in a timely manner; As an embodiment of the present application, the threshold setting includes: Static threshold: set an absolute threshold based on equipment safety specifications, such as a pressure sensor range of 0-100 MPa, with a fault threshold of >90 MPa; Dynamic threshold: generate adaptive threshold through historical data statistical analysis, such as the mean value of the vibration amplitude of a certain device plus 3 times the standard deviation as the early warning threshold, combined with real-time dynamic adjustment of equipment operating conditions, such as allowing the upper limit of vibration amplitude to increase by 20% during high-speed navigation; Three-level early warning system: yellow warning: parameters exceed the normal range by 10-20% (such as temperature > rated value 10%), prompt attention; orange warning: parameters exceed the normal range by 20-50%, suggest scheduling inspection; red alert: parameters exceed the normal range by more than 50% or trigger safety interlock conditions, immediately shut down for maintenance, wherein the safety interlock conditions are usually set based on international maritime standards or equipment manufacturer safety specifications.

[0051] The abnormality detection algorithm includes: Single-class classification algorithm: use One-Class SVM algorithm to establish normal state data boundary, detect data outside the boundary, suitable for scenarios where the normal mode is known; Unsupervised learning: Isolation Forest is used to detect sudden anomalies, such as the amplitude of vibration suddenly rising to 5 times the mean value. Autoencoder is used to compress and reconstruct the data. When the reconstruction error is greater than the threshold, it is determined to be abnormal, suitable for unknown fault mode detection; Time series analysis: ARIMA model is used to predict the trend of normal data. When the deviation between actual value and predicted value is greater than 2 times the standard deviation, an alarm is triggered.

[0052] Specifically, the present application establishes a "digital fingerprint" of the normal operation of the equipment through one-class classification algorithm, accurately identifies the deviation of the abnormal from the known normal mode, and is suitable for equipment with clear known normal state, such as new equipment with stable normal mode after being put into use. For example, monitor the vibration data of newly installed pumps. When the vibration amplitude exceeds the normal range obtained by training, it indicates installation error or loose parts. Through unsupervised learning, unknown types of abnormal patterns are found, solving the problem of "unknown unknown" fault detection. For example, monitor the long-term trend of generator winding temperature. When the reconstruction error continues to increase, it indicates the risk of insulation aging. Through time series analysis, the time correlation of data is captured, the normal trend is predicted, and the deviation is detected. For example, according to historical data, the temperature range of a certain equipment under a certain working condition is predicted, such as 80-90℃ when sailing at full speed. When the actual temperature continues to be lower than 80℃, it indicates that the cooling system is abnormal.

[0053] The maintenance suggestion module in the present application diagnoses the fault type of the equipment according to the output results of the data analysis module, determines the fault cause and influence range, generates specific maintenance operation suggestions, including maintenance steps, required tools and spare parts, etc. Combined with the running state of the equipment and maintenance resources, the maintenance plan is optimized, and the maintenance time and resources are reasonably arranged.

[0054] Specifically, if the vibration amplitude of the main engine bearing of the ship continuously exceeds the orange warning threshold (>8g), the data analysis module diagnoses it as "bearing outer ring wear", and the maintenance operation suggestion generated by the maintenance suggestion module is as follows: Preparation stage: Tools: torque wrench (200-500N・m), bearing puller, clean cloth, lubricating oil (model: Shell Tellus S2 MX 46); Spare parts: bearing assembly (model: SKF 6208-2RS1), sealing ring (size: 40x68x10mm).

[0055] Execution steps: ① Shutdown and lock, disconnect power; ② Disassemble the bearing end cover (8 M12 bolts, torque 80N・m), record the installation position of each part; ③ Use the puller to remove the old bearing, check the outer ring raceway wear condition (wear depth > 0.5mm need to replace); ④ Clean the bearing seat, apply a thin layer of lubricating oil, install new bearings (note the installation direction, the identification surface faces outward); ⑤ Reset the end cover, tighten the bolts in diagonal order, and check the torque to the standard value; ⑥ Start the trial operation, monitor the vibration amplitude < 5g and temperature < 70℃ as the repair is completed.

[0056] Maintenance plan optimization: Combined with the remaining useful life prediction of the bearing (RUL=300 hours), the next port stop time (48 hours later) and the spare parts inventory, the maintenance task is arranged during the port stop period to avoid downtime during the voyage, and it is expected to reduce the downtime by 0.5 days.

[0057] The user interface module in the application provides a user with an intuitive operation interface, including equipment state monitoring, historical record query and maintenance operation guidance function, equipment state monitoring: real-time display of equipment running state, including equipment health condition, fault alarm, etc.; historical record query: provides historical running data and maintenance record of equipment, facilitates crew to check equipment usage and maintenance history; maintenance operation guidance: provides detailed steps and guidance of maintenance operation, helps crew to quickly complete maintenance task.

[0058] The security management module in the application performs encryption processing on the stored and transmitted data to prevent data from being illegally obtained and tampered with; a strict access permission system is set to ensure that only authorized personnel can access system data and perform key operations; the running state of the system is monitored in real time to timely discover and handle system failures and abnormal situations.

[0059] As an embodiment of the application, the encryption processing includes: Transport layer encryption: TLS1.3 protocol is used to encrypt data transmission between sensors and edge nodes, edge nodes and cloud, key exchange uses ECDHE algorithm (Elliptic Curve Diffie-Hellman), data encryption uses AES-256-GCM (Authenticated Encryption Mode, provides integrity check), ensures that the transmitted data cannot be tampered with; Storage layer encryption: AES-256-CBC mode is used to encrypt sensitive fields (such as spare parts inventory quantity, maintenance cost amount), the key is generated and managed by hardware security module (HSM), and is replaced regularly (updated once a week), homomorphic encryption technology is used to encrypt database backup files, supports retrieval and statistics in ciphertext state (such as querying "device list with temperature > 80℃" without decryption).

[0060] As an embodiment of the present application, the access permission system authorizes different data access permissions and operation permissions according to different roles based on the RBAC model, and the access permission system is shown in Table 2:

[0061] Table 2 Access permission system The access permission system is based on the RBAC (role-based access control) model, supports permission subdivision according to device type (such as only power equipment engineers can view host data), time range (such as sensitive data cannot be exported at night), operation risk level (double-approval is required for deleting data), and key operation logs are stored in real time on the chain for evidence (blockchain is tamper-proof), ensuring the security of system data and the stable operation of the system, preventing data leakage and system failure.

[0062] The present application also provides a ship equipment management method based on artificial intelligence, which adopts the above-mentioned ship equipment management system based on artificial intelligence, and specifically includes the following steps: S1: Real-time collection of running data of each equipment on the ship through multiple types of sensors; S2: Storage and management of collected equipment data; S3: Analysis and prediction of equipment data by using machine learning algorithm; S4: Providing specific maintenance operation suggestions and maintenance plans according to the analysis results; S5: Real-time display of the running state, historical running data, maintenance records and maintenance operation guidance of the equipment through the user interface; S6: Encryption processing of the stored and transmitted data, and setting of an access permission system.

[0063] The above description is only some preferred embodiments of the present disclosure and an explanation of the applied technical principles. Those skilled in the art should understand that the scope of the application involved in the embodiments of the present disclosure is not limited to the technical solutions formed by the specific combinations of the above technical features, and should also cover other technical solutions formed by any combination of the above technical features or equivalent features without departing from the above inventive concept. For example, the above features are replaced with the technical features disclosed in the embodiments of the present disclosure (but not limited to) having similar functions to form technical solutions.

Claims

1. A ship equipment management system based on artificial intelligence, characterized in that: include: The data acquisition module collects the operating data of various equipment on board in real time through multiple types of sensors; A data storage module, used for storing and managing the equipment data collected by the data collection module; Data analysis module, which uses machine learning algorithms to analyze and predict equipment data; A maintenance suggestion module, configured to provide specific maintenance operation suggestions and maintenance plans based on the analysis results of the data analysis module; User interface module, used to provide an intuitive operating interface, real-time display of equipment operating status, and provide historical equipment operating data, maintenance records, and repair operation instructions; The security management module ensures the security of system data and the stable operation of the system by encrypting the stored and transmitted data and setting up an access permission system.

2. The artificial intelligence-based ship equipment management system according to claim 1, characterized in that: The data acquisition module includes: Multi-type sensor system: Multiple types of sensors are installed according to the classification of ship equipment to monitor the operating status of the equipment; Data acquisition unit: aggregates and performs edge computing preprocessing on equipment data collected by multiple types of sensor systems; Data transmission unit: transmits equipment data to the data storage module through wireless communication technology.

3. The ship equipment management system based on artificial intelligence according to claim 2, characterized in that: The multi-type sensor system includes: Vibration sensor: used to monitor equipment vibration amplitude and frequency; Temperature sensor: used to monitor equipment temperature changes; Pressure sensor: used to monitor equipment pressure changes; Humidity sensor: used to monitor changes in ambient humidity; Displacement sensor: used to monitor the gap changes between components; Strain sensors: used to monitor structural stress; Speed ​​sensor: used to monitor motor speed; Current sensor: used to monitor load current fluctuations; Flow sensor: used to monitor pipeline medium flow.

4. The ship equipment management system based on artificial intelligence according to claim 1, characterized in that: The data storage module includes: Structured database: used to store basic equipment information and maintenance records. The basic equipment information includes equipment model, manufacturer, installation location, and technical parameters; Distributed storage unit: used to store equipment operation data and achieve horizontal expansion through distributed architecture.

5. The artificial intelligence-based ship equipment management system according to claim 2, characterized in that: The data analysis module includes: Data preprocessing module: cleans and preprocesses the equipment data after edge computing preprocessing to remove noise data and outliers; Feature extraction module: used to extract relevant features from the equipment data processed by the data preprocessing module, including equipment operating time, vibration frequency, and temperature changes; Machine learning modeling: Different machine learning algorithms are used for different task types to model the equipment data after feature extraction to predict the health status and failure risk of the equipment; Abnormality detection module: By setting thresholds and abnormality detection algorithms, it can identify abnormal conditions in equipment operation and issue alarms in a timely manner.

6. The artificial intelligence-based ship equipment management system according to claim 5, characterized in that: The different task types include fault classification, multi-classification, life prediction, and image recognition, and the corresponding machine learning algorithms include support vector machines, random forests, long-term segment memory networks, and convolutional neural networks.

7. The artificial intelligence-based ship equipment management system according to claim 5, characterized in that: The threshold setting includes: Static threshold: Set an absolute threshold based on device security specifications; Dynamic threshold: Generate adaptive threshold through statistical analysis of historical data; Three-level early warning system: Yellow warning: Parameters exceed the normal range by 10%-20%, prompting attention; Orange warning: Parameters exceed the normal range by 20%-50%, and inspections are arranged; Red alert: Parameters exceed the normal range by more than 50% or trigger safety interlock conditions, and the machine is immediately shut down for maintenance.

8. The artificial intelligence-based ship equipment management system according to claim 1, characterized in that: The encryption process includes: Transport layer encryption: TLS 1.3 is used to encrypt data transmission between sensors and edge nodes, and between edge nodes and the cloud. The ECDHE algorithm is used for key exchange, and AES-256-GCM is used for data encryption. Storage layer encryption: AES-256-CBC mode is used to encrypt sensitive fields, and homomorphic encryption technology is used to encrypt database backup files.

9. The artificial intelligence-based ship equipment management system according to claim 1, characterized in that: The access permission system authorizes different data access permissions and operation permissions according to different roles based on the RBAC model.

10. A ship equipment management method based on artificial intelligence, characterized in that: The artificial intelligence-based ship equipment management system according to any one of claims 1 to 9 specifically comprises the following steps: S1: Collect the operating data of various equipment on board in real time through multiple types of sensors; S2: Store and manage the collected equipment data; S3: Use machine learning algorithms to analyze and predict equipment data; S4: Provide specific repair operation suggestions and maintenance plans based on the analysis results; S5: Display the equipment's operating status, historical operating data, maintenance records, and repair operation instructions in real time through the user interface; S6: Encrypt stored and transmitted data and set up an access permission system.

Images