Intelligent ship fault diagnosis system and method based on big data

Abstract

The invention discloses a steamship intelligent fault diagnosis system and method based on big data, and the system comprises a data collection module, a data transmission module, a data storage and processing module, a fault diagnosis model building module, a diagnosis and early warning module, a visual interaction module and the like, data collection covers multiple types of data, the collection frequency and precision are required, and the fault diagnosis model employs a deep learning algorithm and the like. The system has a self-adaptive learning capability, diagnosis and early warning are accurate and timely, visual interaction facilitates user operation and remote access, the system further comprises innovative parts such as an energy efficiency analysis and spare part management module, and all the modules work cooperatively to ensure high efficiency and intelligence of ship fault diagnosis. Through cooperative work of multiple modules, accurate acquisition and analysis of steamship operation data and intelligent fault diagnosis and early warning are realized, and the system also has the functions of energy efficiency analysis and spare part management. The method improves the fault diagnosis accuracy and timeliness, reduces the energy consumption, guarantees the safe and efficient operation of a ship, and improves the operation management level.

Description

Technical Field

[0001] The present invention relates to the technical field of ship fault diagnosis systems, and in particular to a ship intelligent fault diagnosis system and method based on big data. Background Art

[0002] With the booming global shipping industry, ships play a vital role in international trade and personnel transportation. However, as a complex large-scale system, ships face many challenges in their operation, among which timely diagnosis and handling of faults are key issues to ensure safe and efficient operation of ships.

[0003] Traditional ship fault diagnosis methods mainly rely on manual experience and regular maintenance. Maintenance personnel determine the problem by regularly inspecting and maintaining ship equipment and conducting on-site inspections after a fault occurs. This method has many limitations:

[0004] First, manual inspections are difficult to achieve real-time monitoring of the ship's operating status. During the voyage, the ship's various systems and components are in continuous operation, and failures may occur at any time. However, manual inspections can often only be carried out at specific times, and cannot capture instantaneous signs of failure or abnormal conditions in time. For example, some temporary overheating or abnormal vibrations of the engine may be missed if they are not detected at the time of occurrence, resulting in further deterioration of the failure and affecting the normal operation and safety of the ship.

[0005] Secondly, the limitations of manual experience lead to low accuracy of fault diagnosis. Ship systems and equipment are becoming increasingly complex, involving technologies in multiple fields such as mechanics, electronics, electrical, and hydraulics. Different types of faults may show similar symptoms, and it is difficult to accurately determine the root cause of the fault based on manual experience alone. In addition, differences in the technical level and experience of maintenance personnel will also affect the consistency and reliability of diagnostic results. For some new and complex faults, traditional empirical methods may not be able to effectively respond, and misdiagnosis or missed diagnosis may occur easily.

[0006] Furthermore, the regular maintenance method lacks pertinence, is inefficient and has high costs. Regular maintenance usually involves comprehensive inspection and maintenance of ship equipment at fixed time intervals. Regardless of whether the equipment really needs maintenance, a series of operations must be performed. This not only wastes a lot of time and manpower and material resources, but may also cause certain damage to the equipment itself and shorten its service life due to frequent disassembly and assembly of equipment. At the same time, for some sudden failures that occur between two maintenance intervals, regular maintenance cannot be handled in time, which may lead to serious consequences.

[0007] In terms of data processing, traditional methods lack effective utilization of massive operating data. With the development of sensor technology and information technology, a large number of sensors are installed on ships, which can collect various operating data in real time, such as power system parameters, navigation data, hull structure status, etc. However, these data are often simply recorded and stored without in-depth analysis and mining. Traditional data processing methods cannot extract valuable information and fault characteristics from these massive data, and it is difficult to discover potential fault hazards and patterns.

[0008] In addition, the operating environment of ships is complex and changeable, and is affected by many factors such as ocean climate and sea conditions. For example, severe sea conditions may cause the hull to shake violently, which will cause additional stress and interference to the operation of the equipment and increase the probability of failure. Traditional fault diagnosis methods are often not comprehensive enough when considering these complex environmental factors, and it is difficult to accurately assess the impact of environmental factors on ship failures.

[0009] In summary, traditional ship fault diagnosis methods can no longer meet the requirements of modern shipping industry for ship safety, reliability and efficient operation. Therefore, there is an urgent need for an intelligent fault diagnosis system and method based on big data, which can monitor the ship operation status in real time and accurately, diagnose faults quickly and accurately, and provide effective early warnings and solutions to improve the operation management level and safety of ships, reduce maintenance costs and failure rates, and ensure the smooth development of the shipping industry. Summary of the invention

[0010] The present invention proposes a ship intelligent fault diagnosis system and method based on big data to solve the problems mentioned in the above-mentioned prior art.

[0011] In order to achieve the above object, the present invention adopts the following technical solution: a ship intelligent fault diagnosis system based on big data, comprising:

[0012] The data acquisition module is used to collect various data in real time during the operation of the ship, including but not limited to ship power system data, navigation system data, communication system data, hull structure data and environmental data. The power system data covers engine speed, oil temperature, water temperature, oil pressure, and torque parameters. The navigation system data includes GPS positioning information, heading, speed, and rudder angle. The communication system data includes signal strength, frequency, and bit error rate. The hull structure data includes hull stress, deformation, and vibration frequency. Environmental data includes wind speed, wind direction, wave height, sea water temperature, and salinity. The data collection frequency is not less than 10 times / second to ensure real-time and accurate monitoring of the ship's operating status. The accuracy of data collection meets industry standards, and the measurement error of key parameters does not exceed ±0.5%;

[0013] In data collection, the signal data collected by the sensor is preprocessed using the following data fusion formula:

[0014]

[0015] Where S f is the fused data signal, S i is the data signal collected by the i-th sensor, w i is the weight coefficient of the i-th sensor data, which is determined according to the accuracy and reliability factors of the sensor, and Through this data fusion method, the accuracy and reliability of the data can be improved.

[0016] The data transmission module transmits the collected data to the data storage and processing center in real time through wired or wireless transmission. A reliable transmission protocol is used to ensure the integrity and accuracy of data transmission, and the data transmission success rate is not less than 98%. In the case of wireless transmission, it has signal enhancement and anti-interference technology to adapt to the complex operating environment of the ship and ensure stable data transmission under harsh conditions. It supports multiple transmission interfaces, such as Ethernet interface, serial port, and wireless network card, which is convenient for connecting with different types of data sources and data storage and processing centers;

[0017] The data storage and processing center includes a data storage unit and a big data processing unit. The data storage unit uses high-performance storage devices and can store at least 1 year of historical ship operation data. The storage capacity can be expanded according to actual needs. The big data processing unit uses a distributed computing framework and parallel processing technology to quickly process and analyze massive amounts of ship operation data. It has data cleaning, denoising, and fusion functions, and can remove outliers and noise in the data to improve data quality. Data compression technology is used to compress the stored data to save storage space. The data compression ratio is not less than 3:1, and the accuracy and integrity of the data after decompression are guaranteed;

[0018] The fault diagnosis model building module builds a ship fault diagnosis model based on big data analysis technology and machine learning algorithms. It can automatically learn fault characteristics and patterns from historical fault data and normal operation data, and use deep learning algorithms such as convolutional neural network (CNN) and recurrent neural network (RNN) to build a deep neural network model. The model has adaptive learning capabilities and can automatically update and optimize model parameters as new data continues to flow in to improve the accuracy and adaptability of fault diagnosis. During the model training process, cross-validation and regularization techniques are used to prevent model overfitting, and the model training accuracy is not less than 90% on the validation set;

[0019] In the fault diagnosis model building module, the formulas for evaluating the performance of the model and optimizing the model parameters are introduced:

[0020] E=α×A+β×P+γ×F

[0021] Among them: E represents the comprehensive evaluation index of the model, A represents the accuracy index, which can be expressed by the accuracy of the current model on the test set, that is, the ratio of the number of correctly diagnosed samples to the total number of samples. For example, if there are N samples in the test set, of which M are correctly diagnosed, then A = M / N.

[0022] P represents the precision index. For a specific fault type, the precision refers to the proportion of samples that actually belong to this fault type among all samples diagnosed as this fault type. The calculation formula is TP represents true positive examples (i.e., the number of samples that are actually of this fault type and are correctly diagnosed as this fault type), and FP represents false positive examples (i.e., the number of samples that are actually not of this fault type but are diagnosed as this fault type).

[0023] F represents the recall rate indicator. For a specific fault type, the recall rate refers to the proportion of samples that are correctly diagnosed as this fault type among all samples that actually belong to this fault type. The calculation formula is Where FN represents false negative examples (i.e., the number of samples that are actually of this fault type but are mistakenly diagnosed as other types or normal).

[0024] α, β, and γ are weight coefficients of accuracy, precision, and recall, respectively, which are set according to actual application requirements, and α+β+γ=1.

[0025] This formula can comprehensively consider the accuracy, precision and recall of the model, and provide a more comprehensive evaluation index for model optimization. For example, when diagnosing faults in the key power system of a ship, if the tolerance for false alarms is low, the weight coefficient β of the precision rate can be appropriately increased; if you want to detect as many real faults as possible to avoid missed diagnoses, you can increase the weight coefficient γ of the recall rate. At the same time, by continuously adjusting the model parameters and optimizing the algorithm, the comprehensive evaluation index is optimized, thereby improving the performance and reliability of the fault diagnosis model. ;

[0026] The fault diagnosis and early warning module uses the constructed fault diagnosis model to analyze and diagnose the real-time collected data, monitor the ship's operating status in real time, and issue a fault warning in time when fault signs or abnormalities are detected. It can accurately identify various types of faults, including but not limited to power system faults, navigation system faults, communication system faults, and hull structure faults, and the accuracy of fault diagnosis is not less than 85%. Early warning methods include sound and light alarms, SMS notifications, and system pop-ups to ensure that relevant personnel can receive fault information in a timely manner. At the same time, a detailed fault diagnosis report is provided, which includes the fault type, fault location, possible cause analysis, and recommended treatment measures;

[0027] In fault diagnosis, for power system fault diagnosis, the following fault probability evaluation formula is used:

[0028]

[0029] Where P f is the probability of power system failure, x 1 ...x n Monitoring parameters of the power system, including oil temperature, speed, a n and b are the model parameters obtained through historical data training. f When the set threshold is exceeded, it is determined that there may be a fault in the power system, and further detailed diagnosis and early warning are performed.

[0030] The visualization interaction module provides users with an intuitive visualization interface to display the ship's operating status, fault diagnosis results, and data analysis chart information. Users can use this interface to view the ship's operating parameters and health status in real time, and query and analyze fault diagnosis results. It supports users to customize display content and layout to meet the personalized needs of different users. With interactive operation functions, users can use the interface to set parameters, query data, and manage models. The operation response time does not exceed 1 second, providing a good user experience.

[0031] The energy efficiency analysis module monitors the energy consumption of the ship in real time, including fuel consumption and electricity consumption. By analyzing the power system operation data, navigation data and environmental data, the energy efficiency of the ship under different operating conditions is evaluated. The energy efficiency evaluation model is used to calculate the energy utilization index and compare and analyze it with historical data and industry standards. It can identify abnormal energy consumption, such as fuel leakage and excessive equipment energy consumption. For example, the energy utilization rate is calculated by the following formula:

[0032]

[0033] Where η is the energy utilization rate, E o is the effective output energy (such as the propulsion power of a ship), E i To input energy (such as the chemical energy of fuel). When the energy utilization rate deviates from the normal range, the system will issue an energy efficiency abnormality prompt to help users discover energy consumption problems in time, take corresponding measures to reduce energy consumption, and improve the operating economy of the ship.

[0034] The spare parts management module predicts and manages the demand for spare parts based on the ship's equipment model, historical fault data, and the prediction results of the fault diagnosis model. A spare parts inventory model is established to monitor the inventory quantity and usage of spare parts in real time. When the inventory is lower than the set threshold, a procurement suggestion is automatically generated to remind relevant personnel to replenish spare parts in a timely manner. At the same time, combined with the ship's location information and logistics network, the distribution path of spare parts is optimized to ensure that the required spare parts can be obtained in time when needed, reducing downtime caused by waiting for spare parts. For example, the following inventory forecasting formula is used:

[0035] Q t =Q 0 +D t -S t

[0036] Where Q t is the predicted inventory quantity of spare parts at time t, Q 0 is the current inventory quantity, D t is the expected demand quantity in time period t (calculated based on historical data and fault prediction model), S t is the estimated supply quantity within time period t (considering the procurement cycle and logistics time). Through this formula, the system can reasonably plan the inventory of spare parts and improve the maintenance and support capabilities of the ship.

[0037] Furthermore, the data acquisition module also includes sensor self-diagnosis and calibration functions, which can monitor the working status of the sensor in real time, and automatically calibrate or issue a fault prompt when the sensor fails or the data deviates. The accuracy of the sensor self-diagnosis is not less than 95%, and the measurement error of the calibrated sensor does not exceed ±0.3%.

[0038] Furthermore, the fault diagnosis model building module supports multiple model fusion technologies, and can fuse different types of machine learning models, such as combining rule-based models with deep learning models, to improve the accuracy and reliability of fault diagnosis. The fault diagnosis accuracy of the fused model is not less than 5% higher than that of a single model on the validation set.

[0039] Furthermore, the fault diagnosis and warning module has a fault prediction function. Based on historical data and real-time data analysis, it predicts possible faults that may occur in the ship in the future, issues early warnings in advance, and provides reference for repair and maintenance plans. The time for fault prediction is not less than 24 hours in advance, and the prediction accuracy is not less than 70%.

[0040] Furthermore, the visual interaction module supports remote access and mobile terminal access, and users can view the ship's operating status and fault diagnosis results in real time on a remote computer or mobile device via the Internet. The response time of remote access and mobile terminal access does not exceed 3 seconds, ensuring that users can obtain information in a timely manner, which is convenient for remote monitoring and management of ships.

[0041] Furthermore, a method for applying a ship intelligent fault diagnosis system based on big data comprises the following steps:

[0042] S1. Data collection: various sensors and monitoring equipment installed on the ship are used to collect power system data, navigation system data, communication system data, hull structure data and environmental data in real time during the operation of the ship at a frequency of not less than 10 times per second, ensuring that the accuracy of data collection meets the requirements of industry standards and the measurement error of key parameters does not exceed ±0.5%;

[0043] S2. Data transmission: The collected data is transmitted to the data storage and processing center in real time through wired or wireless transmission, using a reliable transmission protocol to ensure the integrity and accuracy of data transmission. The data transmission success rate is not less than 98%;

[0044] S3, data storage and processing, in the data storage and processing center, the transmitted data is stored and processed. The data storage unit uses high-performance storage devices to store at least 1 year of historical ship operation data, and uses data compression technology to compress the data, with a data compression ratio of no less than 3:1. The big data processing unit uses a distributed computing framework and parallel processing technology to clean, denoise, and fuse the data to remove outliers and noise in the data and improve data quality;

[0045] S4. Fault diagnosis model construction: Based on big data analysis technology and machine learning algorithms, fault features and patterns are automatically learned from historical fault data and normal operation data to build a ship fault diagnosis model. Deep learning algorithms, such as convolutional neural network (CNN) and recurrent neural network (RNN), are used to build a deep neural network model. Cross-validation and regularization techniques are used during model training to prevent model overfitting and ensure that the model training accuracy is not less than 90% on the validation set. The model has adaptive learning capabilities and can automatically update and optimize model parameters as new data continues to flow in.

[0046] S5. Fault diagnosis and early warning: Use the constructed fault diagnosis model to analyze and diagnose the real-time collected data and monitor the ship's operating status in real time. When fault signs or abnormalities are detected, timely fault warnings are issued to accurately identify various types of faults, with an accuracy rate of fault diagnosis of no less than 85%. Early warning methods include sound and light alarms, SMS notifications, and system pop-ups. At the same time, a detailed fault diagnosis report is provided, which includes the fault type, fault location, possible cause analysis, and recommended treatment measures.

[0047] In the fault diagnosis and early warning step, the following time series prediction model is used for fault prediction:

[0048]

[0049] where y t+h is the predicted fault index value at time t+h (such as the abnormality of a key parameter), μ is a constant term, and θ j are model parameters, p and q are the order of the model, y t-i is the fault index value at the past i moments, ∈ t-j is the prediction error of the past j moments, ∈ t+h is the prediction error at time t+h. Through learning from historical data and parameter estimation, the model can predict the development of faults in the future based on the current and past fault data trends, providing a basis for early warning.

[0050] S6. Visual display: Through the visual interaction module, it provides users with an intuitive visual interface to display the ship's operating status, fault diagnosis results, and data analysis chart information. Users can use this interface to view the ship's operating parameters and health status in real time, and query and analyze fault diagnosis results. It supports user-defined display content and layout, and has interactive operation functions. Users can use the interface to set parameters, query data, and manage models, and the operation response time does not exceed 1 second.

[0051] S7. Energy efficiency analysis: Real-time calculation of the ship's energy utilization index. Based on the power system operation data, navigation data, and environmental data, the energy utilization calculation formula is used for analysis. By comparing historical data with industry standards, when energy efficiency is abnormal, prompts are issued and cause analysis is performed to provide data support for the formulation of energy-saving measures. For example, energy consumption patterns under different navigation speeds and loads are analyzed to find the best operating conditions to improve the ship's energy efficiency and reduce operating costs.

[0052] S8. Spare parts management: Based on the ship's equipment information, historical fault data, and the prediction results of the fault diagnosis model, the spare parts inventory prediction formula is used to predict the demand for spare parts. The inventory quantity is monitored in real time. When the inventory is below the threshold, procurement suggestions are automatically generated and the delivery route is optimized. Combined with the ship's location and logistics information, it ensures that spare parts can be obtained in time when needed, reducing downtime and improving the ship's maintenance and support capabilities.

[0053] Furthermore, the data collection step also includes a sensor self-diagnosis and calibration step to monitor the working status of the sensor in real time. When the sensor fails or the data deviates, it automatically calibrates or issues a fault prompt. The accuracy of the sensor self-diagnosis is not less than 95%, and the measurement error of the calibrated sensor does not exceed ±0.3%.

[0054] Furthermore, in the fault diagnosis model construction step, multiple model fusion technologies are supported to fuse different types of machine learning models, such as combining rule-based models with deep learning models, to improve the accuracy and reliability of fault diagnosis. The fault diagnosis accuracy of the fused model is not less than 5% higher than that of a single model on the validation set.

[0055] Furthermore, in the fault diagnosis and early warning step, there is also a fault prediction step, which predicts the possible faults of the ship in the future based on historical data and real-time data analysis, issues early warnings in advance, and provides reference for repair and maintenance plans. The time for fault prediction is not less than 24 hours in advance, and the prediction accuracy is not less than 70%.

[0056] Furthermore, in the visualization step, remote access and mobile terminal access are supported, and users can view the ship's operating status and fault diagnosis results in real time on a remote computer or mobile device via the Internet. The response time of remote access and mobile terminal access does not exceed 3 seconds, which facilitates remote monitoring and management of the ship.

[0057] Compared with the prior art, the present invention has the following beneficial effects:

[0058] The present invention uses a high-precision data acquisition module to collect comprehensive ship operation data at a frequency of no less than 10 times per second, ensuring real-time and accurate monitoring of the ship's operating status, with errors controlled within a very small range, and can promptly discover potential problems. The application of data fusion formulas improves data quality and provides a more reliable basis for subsequent analysis.

[0059] The innovative fault diagnosis model building module uses deep learning algorithms to build deep neural network models, which can automatically learn fault characteristics and patterns. The adaptive learning ability enables it to continuously optimize parameters with new data. Combined with cross-validation and regularization technology, the model training accuracy is not less than 90% on the validation set, which greatly improves the accuracy and adaptability of fault diagnosis, and can accurately identify various types of faults with a diagnostic accuracy of not less than 85%.

[0060] The fault diagnosis and early warning module can not only detect fault signs in time and issue early warnings in various ways, but also provide detailed diagnostic reports, which facilitate relevant personnel to quickly understand the fault situation and take measures. The fault prediction function predicts possible faults at least 24 hours in advance, provides a reference for maintenance plans, and effectively reduces the losses caused by sudden faults.

[0061] The visual interaction module provides an intuitive interface, supports remote access and mobile terminal access, has a short response time, and is convenient for users to view the ship status and diagnostic results in real time, perform personalized settings and interactive operations, and realize convenient remote monitoring and management.

[0062] The energy efficiency analysis module calculates indicators such as energy utilization rate to identify energy consumption anomalies, help optimize ship operating conditions, reduce energy consumption, and improve operational economy. The spare parts management module ensures timely acquisition of required spare parts based on accurate demand forecasting and inventory management, reduces downtime, and improves maintenance support capabilities. In short, the present invention comprehensively improves the efficiency and quality of ship fault diagnosis and ensures the safe and efficient operation of ships. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 This is a schematic block diagram of the big data-based ship intelligent fault diagnosis system proposed in the present invention. DETAILED DESCRIPTION

[0064] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0065] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "clockwise" and "counterclockwise" indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the present invention and simplifying the description, and do not indicate or imply that the referred device or element must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.

[0066] In addition, the terms "first" and "second" are used for descriptive purposes only and cannot be understood as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Therefore, the features defined as "first" and "second" may explicitly or implicitly include one or more of the features. In the description of the present invention, the meaning of "multiple" is two or more, unless otherwise clearly and specifically defined. In addition, the terms "installed", "connected" and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or it can be indirectly connected through an intermediate medium, and it can be the internal connection of two elements. For ordinary technicians in this field, the specific meanings of the above terms in the present invention can be understood according to specific circumstances. The present invention will be further described in detail below in conjunction with the accompanying drawings.

[0067] Reference Figure 1 :A ship intelligent fault diagnosis system based on big data, comprising:

[0068] The data acquisition module is used to collect various data in real time during the operation of the ship, including but not limited to ship power system data, navigation system data, communication system data, hull structure data and environmental data. The power system data covers engine speed, oil temperature, water temperature, oil pressure, and torque parameters. The navigation system data includes GPS positioning information, heading, speed, and rudder angle. The communication system data includes signal strength, frequency, and bit error rate. The hull structure data includes hull stress, deformation, and vibration frequency. Environmental data includes wind speed, wind direction, wave height, sea water temperature, and salinity. The data collection frequency is not less than 10 times / second to ensure real-time and accurate monitoring of the ship's operating status. The accuracy of data collection meets industry standards, and the measurement error of key parameters does not exceed ±0.5%;

[0069] In data collection, the signal data collected by the sensor is preprocessed using the following data fusion formula:

[0070]

[0071] Where S f is the fused data signal, S i is the data signal collected by the i-th sensor, w i is the weight coefficient of the i-th sensor data, which is determined according to the accuracy and reliability factors of the sensor, and Through this data fusion method, the accuracy and reliability of the data can be improved.

[0072] The data transmission module transmits the collected data to the data storage and processing center in real time through wired or wireless transmission. A reliable transmission protocol is used to ensure the integrity and accuracy of data transmission, and the data transmission success rate is not less than 98%. In the case of wireless transmission, it has signal enhancement and anti-interference technology to adapt to the complex operating environment of the ship and ensure stable data transmission under harsh conditions. It supports multiple transmission interfaces, such as Ethernet interface, serial port, and wireless network card, which is convenient for connecting with different types of data sources and data storage and processing centers;

[0073] The data storage and processing center includes a data storage unit and a big data processing unit. The data storage unit uses high-performance storage devices and can store at least 1 year of historical ship operation data. The storage capacity can be expanded according to actual needs. The big data processing unit uses a distributed computing framework and parallel processing technology to quickly process and analyze massive amounts of ship operation data. It has data cleaning, denoising, and fusion functions, and can remove outliers and noise in the data to improve data quality. Data compression technology is used to compress the stored data to save storage space. The data compression ratio is not less than 3:1, and the accuracy and integrity of the data after decompression are guaranteed;

[0074] The fault diagnosis model building module builds a ship fault diagnosis model based on big data analysis technology and machine learning algorithms. It can automatically learn fault characteristics and patterns from historical fault data and normal operation data, and use deep learning algorithms such as convolutional neural network (CNN) and recurrent neural network (RNN) to build a deep neural network model. The model has adaptive learning capabilities and can automatically update and optimize model parameters as new data continues to flow in to improve the accuracy and adaptability of fault diagnosis. During the model training process, cross-validation and regularization techniques are used to prevent model overfitting, and the model training accuracy is not less than 90% on the validation set;

[0075] In the fault diagnosis model building module, the formulas for evaluating the performance of the model and optimizing the model parameters are introduced:

[0076] E=α×A+β×P+γ×F

[0077] Among them: E represents the comprehensive evaluation index of the model, A represents the accuracy index, which can be expressed by the accuracy of the current model on the test set, that is, the ratio of the number of correctly diagnosed samples to the total number of samples. For example, if there are N samples in the test set, of which M are correctly diagnosed, then A = M / N.

[0078] P represents the precision index. For a specific fault type, the precision refers to the proportion of samples that actually belong to this fault type among all samples diagnosed as this fault type. The calculation formula is TP represents true positive examples (i.e., the number of samples that are actually of this fault type and are correctly diagnosed as this fault type), and FP represents false positive examples (i.e., the number of samples that are actually not of this fault type but are diagnosed as this fault type).

[0079] F represents the recall rate indicator. For a specific fault type, the recall rate refers to the proportion of samples that are correctly diagnosed as this fault type among all samples that actually belong to this fault type. The calculation formula is Where FN represents false negative examples (i.e., the number of samples that are actually of this fault type but are mistakenly diagnosed as other types or normal).

[0080] α, β, and γ are weight coefficients of accuracy, precision, and recall, respectively, which are set according to actual application requirements, and α+β+γ=1.

[0081] This formula can comprehensively consider the accuracy, precision and recall of the model, and provide a more comprehensive evaluation index for model optimization. For example, when diagnosing faults in the key power system of a ship, if the tolerance for false alarms is low, the weight coefficient β of the precision rate can be appropriately increased; if you want to detect as many real faults as possible to avoid missed diagnoses, you can increase the weight coefficient γ of the recall rate. At the same time, by continuously adjusting the model parameters and optimizing the algorithm, the comprehensive evaluation index is optimized, thereby improving the performance and reliability of the fault diagnosis model. ;

[0082] The fault diagnosis and early warning module uses the constructed fault diagnosis model to analyze and diagnose the real-time collected data, monitor the ship's operating status in real time, and issue a fault warning in time when fault signs or abnormalities are detected. It can accurately identify various types of faults, including but not limited to power system faults, navigation system faults, communication system faults, and hull structure faults, and the accuracy of fault diagnosis is not less than 85%. Early warning methods include sound and light alarms, SMS notifications, and system pop-ups to ensure that relevant personnel can receive fault information in a timely manner. At the same time, a detailed fault diagnosis report is provided, which includes the fault type, fault location, possible cause analysis, and recommended treatment measures;

[0083] In fault diagnosis, for power system fault diagnosis, the following fault probability evaluation formula is used:

[0084]

[0085] Where P f is the probability of power system failure, x 1 ...x n Monitoring parameters of the power system, including oil temperature, speed, a n and b are the model parameters obtained through historical data training. f When the set threshold is exceeded, it is determined that there may be a fault in the power system, and a detailed diagnosis and early warning in the present invention are performed.

[0086] The visualization interaction module provides users with an intuitive visualization interface to display the ship's operating status, fault diagnosis results, and data analysis chart information. Users can use this interface to view the ship's operating parameters and health status in real time, and query and analyze fault diagnosis results. It supports users to customize display content and layout to meet the personalized needs of different users. With interactive operation functions, users can use the interface to set parameters, query data, and manage models. The operation response time does not exceed 1 second, providing a good user experience.

[0087] The energy efficiency analysis module monitors the energy consumption of the ship in real time, including fuel consumption and electricity consumption. By analyzing the power system operation data, navigation data and environmental data, the energy efficiency of the ship under different operating conditions is evaluated. The energy efficiency evaluation model is used to calculate the energy utilization index and compare and analyze it with historical data and industry standards. It can identify abnormal energy consumption, such as fuel leakage and excessive equipment energy consumption. For example, the energy utilization rate is calculated by the following formula:

[0088]

[0089] Where η is the energy utilization rate, E o is the effective output energy (such as the propulsion power of a ship), E i To input energy (such as the chemical energy of fuel). When the energy utilization rate deviates from the normal range, the system will issue an energy efficiency abnormality prompt to help users discover energy consumption problems in time, take corresponding measures to reduce energy consumption, and improve the operating economy of the ship.

[0090] The spare parts management module predicts and manages the demand for spare parts based on the ship's equipment model, historical fault data, and the prediction results of the fault diagnosis model. A spare parts inventory model is established to monitor the inventory quantity and usage of spare parts in real time. When the inventory is lower than the set threshold, a procurement suggestion is automatically generated to remind relevant personnel to replenish spare parts in a timely manner. At the same time, combined with the ship's location information and logistics network, the distribution path of spare parts is optimized to ensure that the required spare parts can be obtained in time when needed, reducing downtime caused by waiting for spare parts. For example, the following inventory forecasting formula is used:

[0091] Q t =Q 0 +D t -S t

[0092] Where Q t is the predicted inventory quantity of spare parts at time t, Q 0 is the current inventory quantity, D t is the expected demand quantity in time period t (calculated based on historical data and fault prediction model), S t is the estimated supply quantity within time period t (considering the procurement cycle and logistics time). Through this formula, the system can reasonably plan the inventory of spare parts and improve the maintenance and support capabilities of the ship.

[0093] In the present invention, the data acquisition module also includes a sensor self-diagnosis and calibration function, which can monitor the working status of the sensor in real time, and automatically calibrate or issue a fault prompt when the sensor fails or the data deviates. The accuracy of the sensor self-diagnosis is not less than 95%, and the measurement error of the sensor after calibration does not exceed ±0.3%.

[0094] In the present invention, the fault diagnosis model building module supports multiple model fusion technologies and can fuse different types of machine learning models, such as combining rule-based models with deep learning models, to improve the accuracy and reliability of fault diagnosis. The fault diagnosis accuracy after model fusion is not less than 5% higher than that of a single model on the validation set.

[0095] In the present invention, the fault diagnosis and early warning module has a fault prediction function. Based on historical data and real-time data analysis, it predicts possible faults of the ship in the future, issues early warnings in advance, and provides reference for repair and maintenance plans. The time advance for fault prediction is not less than 24 hours, and the prediction accuracy is not less than 70%.

[0096] In the present invention, the visualization interaction module supports remote access and mobile terminal access, and users can view the operation status and fault diagnosis results of the ship in real time on a remote computer or mobile device through the Internet. The response time of remote access and mobile terminal access does not exceed 3 seconds, ensuring that users can obtain information in time, which is convenient for remote monitoring and management of the ship.

[0097] In the present invention, a method for applying a ship intelligent fault diagnosis system based on big data comprises the following steps:

[0098] S1. Data collection: various sensors and monitoring equipment installed on the ship are used to collect power system data, navigation system data, communication system data, hull structure data and environmental data in real time during the operation of the ship at a frequency of not less than 10 times per second, ensuring that the accuracy of data collection meets the requirements of industry standards and the measurement error of key parameters does not exceed ±0.5%;

[0099] S2. Data transmission: The collected data is transmitted to the data storage and processing center in real time through wired or wireless transmission, using a reliable transmission protocol to ensure the integrity and accuracy of data transmission. The data transmission success rate is not less than 98%;

[0100] S3, data storage and processing, in the data storage and processing center, the transmitted data is stored and processed. The data storage unit uses high-performance storage devices to store at least 1 year of historical ship operation data, and uses data compression technology to compress the data, with a data compression ratio of no less than 3:1. The big data processing unit uses a distributed computing framework and parallel processing technology to clean, denoise, and fuse the data to remove outliers and noise in the data and improve data quality;

[0101] S4. Fault diagnosis model construction: Based on big data analysis technology and machine learning algorithms, fault features and patterns are automatically learned from historical fault data and normal operation data to build a ship fault diagnosis model. Deep learning algorithms, such as convolutional neural network (CNN) and recurrent neural network (RNN), are used to build a deep neural network model. Cross-validation and regularization techniques are used during model training to prevent model overfitting and ensure that the model training accuracy is not less than 90% on the validation set. The model has adaptive learning capabilities and can automatically update and optimize model parameters as new data continues to flow in.

[0102] S5. Fault diagnosis and early warning: Use the constructed fault diagnosis model to analyze and diagnose the real-time collected data and monitor the ship's operating status in real time. When fault signs or abnormalities are detected, timely fault warnings are issued to accurately identify various types of faults, with an accuracy rate of fault diagnosis of no less than 85%. Early warning methods include sound and light alarms, SMS notifications, and system pop-ups. At the same time, a detailed fault diagnosis report is provided, which includes the fault type, fault location, possible cause analysis, and recommended treatment measures.

[0103] In the fault diagnosis and early warning step, the following time series prediction model is used for fault prediction:

[0104]

[0105] where y t+h is the predicted fault index value at time t+h (such as the abnormality of a key parameter), μ is a constant term, and θ j are model parameters, p and q are the order of the model, y t-i is the fault index value at the past i moments, ∈ t-j is the prediction error of the past j moments, ∈ t+h is the prediction error at time t+h. Through learning from historical data and parameter estimation, the model can predict the development of faults in the future based on the current and past fault data trends, providing a basis for early warning.

[0106] S6. Visual display: Through the visual interaction module, it provides users with an intuitive visual interface to display the ship's operating status, fault diagnosis results, and data analysis chart information. Users can use this interface to view the ship's operating parameters and health status in real time, and query and analyze fault diagnosis results. It supports user-defined display content and layout, and has interactive operation functions. Users can use the interface to set parameters, query data, and manage models, and the operation response time does not exceed 1 second.

[0107] S7. Energy efficiency analysis: Real-time calculation of the ship's energy utilization index. Based on the power system operation data, navigation data, and environmental data, the energy utilization calculation formula is used for analysis. By comparing historical data with industry standards, when energy efficiency is abnormal, prompts are issued and cause analysis is performed to provide data support for the formulation of energy-saving measures. For example, energy consumption patterns under different navigation speeds and loads are analyzed to find the best operating conditions to improve the ship's energy efficiency and reduce operating costs.

[0108] S8. Spare parts management: Based on the ship's equipment information, historical fault data, and the prediction results of the fault diagnosis model, the spare parts inventory prediction formula is used to predict the demand for spare parts. The inventory quantity is monitored in real time. When the inventory is below the threshold, procurement suggestions are automatically generated and the delivery route is optimized. Combined with the ship's location and logistics information, it ensures that spare parts can be obtained in time when needed, reducing downtime and improving the ship's maintenance and support capabilities.

[0109] In the present invention, the data collection step also includes a sensor self-diagnosis and calibration step, which monitors the working status of the sensor in real time. When the sensor fails or the data deviates, it automatically calibrates or issues a fault prompt. The accuracy of the sensor self-diagnosis is not less than 95%, and the measurement error of the sensor after calibration does not exceed ±0.3%.

[0110] In the present invention, in the step of building a fault diagnosis model, multiple model fusion technologies are supported to fuse different types of machine learning models, such as combining rule-based models with deep learning models, to improve the accuracy and reliability of fault diagnosis. The fault diagnosis accuracy after model fusion is not less than 5% higher than that of a single model on the validation set.

[0111] In the present invention, the fault diagnosis and early warning step also includes a fault prediction step, which predicts possible faults of the ship in the future based on historical data and real-time data analysis, issues early warnings in advance, and provides reference for repair and maintenance plans. The time advance for fault prediction is not less than 24 hours, and the prediction accuracy is not less than 70%.

[0112] In the ship intelligent fault diagnosis system of the present invention, the remote access and mobile terminal access functions of the visual display steps play a vital role, bringing great convenience and efficient monitoring and management experience to users.

[0113] The remote access function breaks the geographical restrictions, allowing users to view the ship's operating status and fault diagnosis results in real time on a remote computer anytime and anywhere through the Internet. No matter where the ship is sailing in the world, as long as there is a network connection, relevant personnel such as shipping company operations managers, technical experts, etc. can know all the conditions of the ship as if they were on the ship. The system ensures stable data transmission through advanced network communication technology. During the data transmission process, efficient data encryption algorithms such as AES are used to ensure data security and prevent data from being stolen or tampered with. Users log in to the specially designed visual interface through a browser. The interface layout is clear and reasonable, showing the ship's real-time location information (accurately presented through GPS positioning), navigation key data such as sailing speed and heading, and detailed parameters of the power system, such as the real-time engine speed, oil temperature, water temperature, oil pressure, etc., so that users can fully understand the operation of the ship. At the same time, the system displays the fault diagnosis results with prominent labels and detailed text descriptions, including the specific type of fault (such as engine overheating fault in power system fault, fuel supply system fault, etc.), the location of the fault (accurate to specific equipment components or areas) and possible cause analysis, so that users can quickly obtain key information and make corresponding decisions.

[0114] One of the key advantages of the remote access function is that the response time does not exceed 3 seconds. In order to achieve this efficient response, the system has been optimized at multiple levels. At the data transmission level, a combination of data compression and streaming media technology is adopted. Data compression technologies such as LZ77 can effectively reduce the amount of data transmission and increase the transmission speed, while streaming media technology ensures the real-time update of data. When the user initiates a remote access request, the system can quickly compress and transmit the latest operating data and diagnostic results of the ship, and display them on the user's remote computer screen in a short time. On the server side, it is equipped with high-performance server hardware, such as multi-core processors, large-capacity high-speed memory and high-speed storage devices, and adopts a distributed computing architecture to distribute data processing tasks to multiple nodes for parallel processing, which greatly improves the system's responsiveness. In addition, the system also uses intelligent caching technology to cache frequently accessed data. When the user requests the same data again, it can be read directly from the cache, further shortening the response time and ensuring that users can obtain the latest information of the ship in a timely manner, just like real-time monitoring on site.

[0115] Considering that remote access involves sensitive ship operation data and security issues, the system has established a complete security protection system. In the network transmission layer, in addition to the above-mentioned data encryption algorithm, VPN (virtual private network) technology is also used to establish a dedicated security channel for data transmission, further enhancing the security of data. When users log in to the remote access interface, they need to undergo strict identity authentication, which can be combined in a variety of ways, such as user name and password, dynamic password, digital certificate, etc., to ensure that only authorized users can access the relevant information of the ship. The system also has access rights management functions, which can be finely divided according to the user's role and responsibilities. For example, senior management of a shipping company may have the authority to view the overall operation profile of the ship and key fault information, while technical personnel on board may have the authority to operate and analyze detailed equipment parameters and fault diagnosis details. Users with different permissions can only see the authorized content when accessing the system, which not only ensures the security of the data, but also meets the actual needs of different users in the process of remote monitoring and management of ships, ensuring the reliability and security of remote monitoring.

[0116] The mobile terminal access function further enhances the convenience and flexibility of ship monitoring for users, allowing users to view the ship's operating status and fault diagnosis results anytime and anywhere through mobile devices such as mobile phones and tablets. This is of great practical value for shipping company managers on business trips, staff waiting for ships to dock at ports, and relevant personnel who need to respond to ship emergencies at any time. Whether users are outside the office, at home, or on mobile transportation, as long as the mobile device is connected to the Internet, they can quickly access the ship's information through a dedicated application (APP). The interface design of the APP fully considers the characteristics of mobile devices and user operating habits, using a simple and intuitive layout and a large font and high-contrast display method to facilitate users to quickly obtain key information in different environments. Users can perform simple and intuitive operations through the touch screen, such as zooming in and out on the map to clearly view the real-time position and navigation track of the ship, sliding the screen to switch to view the operating parameters and fault diagnosis reports of different systems, etc. At the same time, the APP also supports voice command operation. Using voice recognition technology, users only need to say the corresponding commands, such as "check the engine speed" and "query the latest fault records", etc. The system can respond quickly and display relevant information, greatly improving the convenience of operation. Especially when the user is not convenient to operate the mobile device with both hands, the voice command function provides users with a more convenient access method.

[0117] The mobile terminal access function is also equipped with real-time push and reminder services to ensure that users can keep up to date with important ship conditions. When a ship fails or experiences an abnormality, the system will immediately send a push notification to the user's mobile device. The notification content contains detailed fault information, such as the fault type (e.g., sudden engine shutdown fault in power system fault, signal loss fault in navigation system, etc.), the severity of the fault (divided into high, medium and low levels, so that users can quickly judge the degree of urgency), and preliminary response suggestions. This enables users to learn about the ship's situation at the first time and take appropriate measures in a timely manner, such as notifying the ship's personnel for emergency treatment and coordinating relevant maintenance resources. In addition to fault notifications, the system can also regularly push ship operation reports based on the user's personalized settings, such as daily mileage, energy consumption statistics, equipment operating hours, etc., so that users can keep abreast of the ship's daily operating conditions. When the equipment needs regular maintenance, the system will send maintenance reminders to relevant personnel a few days in advance, including maintenance items, estimated maintenance time, and required resources, etc., to ensure that the ship's maintenance work is carried out on time, effectively ensuring the safe and stable operation of the ship, and improving the user's response efficiency and decision-making timeliness to ship operation management.

[0118] The above are only preferred specific implementation modes of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can replace or change the technical solution and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. A ship intelligent fault diagnosis system based on big data, characterized in that: include: The data acquisition module is used to collect various data in the operation process of the ship in real time, including ship power system data, navigation system data, communication system data, hull structure data and environmental data. The power system data includes engine speed, oil temperature, water temperature, oil pressure, and torque parameters. The navigation system data includes GPS positioning information, heading, speed, and rudder angle. The communication system data includes signal strength, frequency, and bit error rate. The hull structure data includes hull stress, deformation, and vibration frequency. The environmental data includes wind speed, wind direction, wave height, sea water temperature, and salinity. In data collection, the signal data collected by the sensor is preprocessed using the following data fusion formula: Where S f is the fused data signal, S i is the data signal collected by the i-th sensor, w i is the weight coefficient of the i-th sensor data, and The data transmission module transmits the collected data to the data storage and processing center in real time through wired or wireless transmission, and supports Ethernet interface, serial port, and wireless network card transmission interface; The data storage and processing center includes a data storage unit and a big data processing unit. The data storage unit uses high-performance storage devices to store at least one year of historical ship operation data. The big data processing unit uses a distributed computing framework and parallel processing technology, and has data cleaning, denoising, and fusion functions. It uses data compression technology to compress the stored data, and the data compression ratio is not less than 3:

1. The fault diagnosis model building module builds a ship fault diagnosis model based on big data analysis technology and machine learning algorithms, automatically learns fault features and patterns from historical fault data and normal operation data, and uses convolutional neural network CNN or recurrent neural network RNN ​​deep learning algorithms to build a deep neural network model; In the fault diagnosis model building module, the formulas for evaluating the performance of the model and optimizing the model parameters are introduced: E=α×A+β×P+γ×F Where: E represents the comprehensive evaluation index of the model, A represents the accuracy index, which can be expressed by the accuracy of the current model on the test set, that is, the ratio of the number of correctly diagnosed samples to the total number of samples; P represents the precision index. For a specific fault type, the precision refers to the proportion of samples that actually belong to this fault type among all samples diagnosed as this fault type. The calculation formula is: Where TP represents true positive examples and FP represents false positive examples; F represents the recall rate indicator. For a specific fault type, the recall rate refers to the proportion of samples that are correctly diagnosed as this fault type among all samples that actually belong to this fault type. The calculation formula is: Where FN represents a false counterexample; α, β, and γ are the weight coefficients of accuracy, precision, and recall, respectively, and α+β+γ=1; The fault diagnosis and early warning module uses the constructed fault diagnosis model to analyze and diagnose the real-time collected data, monitor the ship's operating status in real time, and issue a fault warning when a fault sign or abnormality is detected. The fault types include power system fault, navigation system fault, communication system fault, and hull structure fault; In fault diagnosis, for power system fault diagnosis, the following fault probability evaluation formula is used: Where P f is the probability of power system failure, x1...x n Monitoring parameters of the power system, including oil temperature, speed, a n and b are the model parameters obtained through historical data training. f When the set threshold is exceeded, it is determined that there may be a fault in the power system, and further detailed diagnosis and early warning are carried out; Visual interaction module provides users with an intuitive visual interface to display ship operation status, fault diagnosis results, and data analysis chart information; The energy efficiency analysis module monitors the energy consumption of the ship in real time, including fuel consumption and electricity consumption. By analyzing the power system operation data, navigation data and environmental data, it evaluates the energy efficiency of the ship under different operating conditions. It uses the energy efficiency evaluation model to calculate the energy utilization index, and compares and analyzes it with historical data and industry standards. The energy utilization rate is calculated using the following formula: Where η is the energy utilization rate, E o is the effective output energy, E i To input energy, when the energy utilization rate deviates from the normal range, the system will issue an energy efficiency abnormality prompt to help users discover energy consumption problems in a timely manner; The spare parts management module predicts and manages the demand for spare parts based on the ship's equipment model, historical fault data, and the prediction results of the fault diagnosis model, establishes a spare parts inventory model, and monitors the inventory quantity and usage of spare parts in real time. When the inventory is lower than the set threshold, it automatically generates procurement suggestions to remind relevant personnel to replenish spare parts in time. Combined with the ship's location information and logistics network, it optimizes the distribution path of spare parts and uses the following inventory prediction formula: Q t =Q0+D t -S t Where Q t is the predicted inventory quantity of spare parts at time t, Q0 is the current inventory quantity, and D t is the expected demand quantity in time period t, S t is the expected supply quantity in time period t.

2. The ship intelligent fault diagnosis system based on big data according to claim 1 is characterized in that: The data acquisition module also includes sensor self-diagnosis and calibration functions, which can monitor the working status of the sensor in real time. When the sensor fails or the data deviates, it automatically calibrates or issues a fault prompt. The accuracy of the sensor self-diagnosis is not less than 95%, and the measurement error of the calibrated sensor does not exceed ±0.3%.

3. The ship intelligent fault diagnosis system based on big data according to claim 1 is characterized in that: The fault diagnosis model building module supports multiple model fusion technologies, fusing different types of machine learning models, including combining rule-based models with deep learning models, to improve the accuracy and reliability of fault diagnosis. The fault diagnosis accuracy after model fusion is improved by no less than 5% on the verification set compared with a single model.

4. The ship intelligent fault diagnosis system based on big data according to claim 1 is characterized in that: The fault diagnosis and early warning module has a fault prediction function. Based on historical data and real-time data analysis, it predicts possible faults that may occur in the ship in the future, issues early warnings, and provides reference for repair and maintenance plans. The fault prediction time is no less than 24 hours in advance, and the prediction accuracy is no less than 70%.

5. The ship intelligent fault diagnosis system based on big data according to claim 1 is characterized in that: The visualization interaction module supports remote access and mobile terminal access. Users can view the ship's operating status and fault diagnosis results in real time on a remote computer or mobile device via the Internet. The response time of remote access and mobile terminal access does not exceed 3 seconds, ensuring that users can obtain information in a timely manner and facilitating remote monitoring and management of the ship.

6. A method for applying the ship intelligent fault diagnosis system based on big data according to any one of claims 1 to 5, characterized in that: The following steps are involved: S1. Data collection: various sensors and monitoring equipment installed on the ship are used to collect power system data, navigation system data, communication system data, hull structure data and environmental data in real time during the operation of the ship at a frequency of not less than 10 times per second, ensuring that the accuracy of data collection meets the requirements of industry standards and the measurement error of key parameters does not exceed ±0.5%; S2. Data transmission: The collected data is transmitted to the data storage and processing center in real time through wired or wireless transmission, using a reliable transmission protocol to ensure the integrity and accuracy of data transmission. The data transmission success rate is not less than 98%; S3. Data storage and processing: In the data storage and processing center, the transmitted data is stored and processed. The data storage unit uses high-performance storage devices to store at least 1 year of historical ship operation data, and uses data compression technology to compress the data. The data compression ratio is not less than 3:

1. The big data processing unit uses a distributed computing framework and parallel processing technology to clean, denoise, and fuse the data to remove outliers and noise in the data and improve data quality. S4. Fault diagnosis model construction: Based on big data analysis technology and machine learning algorithms, fault features and patterns are automatically learned from historical fault data and normal operation data to build a ship fault diagnosis model. Convolutional neural network and recurrent neural network deep learning algorithms are used to build a deep neural network model. Cross-validation and regularization techniques are used in the model training process to prevent model overfitting, so that the model training accuracy is not less than 90% on the validation set. The model has adaptive learning capabilities and can automatically update and optimize model parameters as new data continues to flow in. S5. Fault diagnosis and early warning: Use the constructed fault diagnosis model to analyze and diagnose the real-time collected data, monitor the ship's operating status in real time, and issue fault warnings in time when fault signs or abnormalities are detected. Accurately identify various types of faults, with an accuracy rate of fault diagnosis of no less than 85%. Early warning methods include sound and light alarms, SMS notifications, and system pop-ups. At the same time, provide a detailed fault diagnosis report, which includes the fault type, fault location, possible cause analysis, and recommended treatment measures; In the fault diagnosis and early warning step, the following time series prediction model is used for fault prediction: where y t+h is the predicted fault index value at time t+h, μ is a constant term, and θ j are model parameters, p and q are the order of the model, y t-i is the fault index value at the past i moments, ∈ t-j is the prediction error of the past j moments, ∈ t+h is the prediction error at time t+h, obtained through learning from historical data and parameter estimation; S6. Visual display: Through the visual interaction module, it provides users with an intuitive visual interface to display the ship's operating status, fault diagnosis results, and data analysis chart information. Users can use this interface to view the ship's operating parameters and health status in real time, query and analyze fault diagnosis results, support user-defined display content and layout, and have interactive operation functions. Users can use the interface to set parameters, query data, and manage models, and the operation response time does not exceed 1 second; S7. Energy efficiency analysis: Real-time calculation of the ship's energy utilization index. Based on the power system operation data, navigation data and environmental data, the energy utilization calculation formula is used for analysis, and historical data and industry standards are compared. When energy efficiency abnormalities are found, prompts are issued and cause analysis is performed to provide data support for the formulation of energy-saving measures. S8. Spare parts management: Based on the ship’s equipment information, historical fault data, and the prediction results of the fault diagnosis model, the spare parts inventory prediction formula is used to predict the demand for spare parts and monitor the inventory quantity in real time. When the inventory is below the threshold, procurement recommendations are automatically generated and the delivery route is optimized, taking into account the ship’s location and logistics information.

7. The method for intelligent ship fault diagnosis based on big data according to claim 6 is characterized in that: The data collection step also includes sensor self-diagnosis and calibration steps to monitor the working status of the sensor in real time. When the sensor fails or the data deviates, it will automatically calibrate or issue a fault prompt. The accuracy of the sensor self-diagnosis is not less than 95%, and the measurement error of the calibrated sensor does not exceed ±0.3%.

8. The method for intelligent ship fault diagnosis based on big data according to claim 6 is characterized in that: In the fault diagnosis model construction step, model fusion technology is supported to fuse different types of machine learning models and combine rule-based models with deep learning models to improve the accuracy and reliability of fault diagnosis. The fault diagnosis accuracy after model fusion is no less than 5% higher than that of a single model on the verification set.

9. The method for intelligent ship fault diagnosis based on big data according to claim 6 is characterized in that: The fault diagnosis and early warning steps also include a fault prediction step. Based on historical data and real-time data analysis, the possible faults that may occur in the ship in the future are predicted, and early warnings are issued to provide a reference for repair and maintenance plans. The fault prediction time is no less than 24 hours in advance, and the prediction accuracy is no less than 70%.

10. The method for intelligent ship fault diagnosis based on big data according to claim 6, characterized in that: In the visualization step, remote access and mobile terminal access are supported. Users can view the ship's operating status and fault diagnosis results in real time on a remote computer or mobile device via the Internet. The response time of remote access and mobile terminal access does not exceed 3 seconds, which facilitates remote monitoring and management of the ship.

Images