Health Management Methods for Integrated Power Systems of Unmanned Surface Vessels

Abstract

This invention provides a health management method for an unmanned surface vessel's integrated power system. It employs fault prediction-based early warning and system reconfiguration-based fault recovery, enabling fault early warning / pre-location, fault reconfiguration / pre-reconfiguration, and maintenance recommendations. This ensures the healthy operation of the integrated power system during unmanned surface vessel operation. During unmanned surface vessel operation, real-time monitoring of data reflecting the operational information of various devices in the integrated power system is performed. Alarm judgments are issued for each monitored data point. When a monitored data point exceeds a set threshold, a steady-state assessment is conducted. For monitored data determined to be in a non-steady-state state, a trend assessment is further performed. If the trend is normal, it indicates that the integrated power system is operating normally; if the trend is abnormal, fault location is performed, and then fault reconfiguration is carried out based on the fault location results.

Description

Technical Field

[0001] This invention relates to a power system health management method, specifically a method for integrated power system health management of unmanned vessels, belonging to the field of intelligent engine room management technology. Background Technology

[0002] As a complex system, the integrated electric propulsion system of unmanned vessels inevitably experiences various failures when operating in harsh environments, significantly impacting the safety, stability, and reliability of the vessel. For a long time, the maintenance strategies for ship electromechanical systems have relied solely on periodic and reactive maintenance. While periodic maintenance can achieve good results, it often leads to over-maintenance or under-maintenance, resulting in a significant waste of human and material resources or insufficient system reliability. Reactive maintenance, on the other hand, often results in irreversible damage due to the fact that failures have already occurred, consuming enormous human and material resources and yielding less effective results.

[0003] As the core system of an unmanned surface vessel (USV), the integrated electric power system is the foundation and guarantee for its mission execution. Due to its unmanned nature, the two maintenance strategies mentioned above are no longer sufficient to meet its operational requirements. To ensure successful mission execution, reliable health management of the USV's integrated electric power system is necessary to improve the USV's combat effectiveness, endurance, and safety. Summary of the Invention

[0004] In view of this, the present invention provides a health management method for the integrated power system of unmanned vessels, which adopts early warning based on fault prediction and fault recovery based on system reconstruction, and can realize fault early warning / pre-location, fault reconstruction / pre-reconstruction, and maintenance suggestions; thereby ensuring the healthy operation of the integrated power system of unmanned vessels during operation.

[0005] A health management method for the integrated power system of unmanned vessels involves real-time monitoring of data that reflects the operating information of various devices in the integrated power system during the operation of the unmanned vessel.

[0006] It also performs alarm judgments on each monitoring data; when a monitoring data exceeds a set threshold, it performs a steady-state judgment on the monitoring data; if it is determined that the monitoring data is only fluctuating within a set range, it is considered that the monitoring data is in a steady state, indicating that the current integrated power system is operating normally; otherwise, it is considered that the monitoring data is in a non-steady state, and the monitoring data is further sent to the trend judgment module for trend judgment.

[0007] The trend judgment can obtain the predicted data of the current monitoring data within a set time range. Based on the predicted data, it can be determined whether the current monitoring data is a normal trend or an abnormal trend within the set time range. If it is a normal trend, it indicates that the current integrated power system is operating normally. If it is an abnormal trend, fault location is performed, and then fault reconstruction is performed based on the fault location results.

[0008] In a preferred embodiment of the present invention, a tiered alarm method is adopted when judging alarms based on monitoring data. Two alarm thresholds are preset for each monitoring data: a shutdown alarm threshold and a reduced operating condition alarm threshold.

[0009] When all monitoring data are less than the corresponding de-operation alarm threshold, it indicates that the current integrated power system is operating normally;

[0010] When a certain monitoring data is greater than the corresponding de-operation alarm threshold but less than the corresponding shutdown alarm threshold, a steady-state judgment is made on the monitoring data.

[0011] When a certain monitoring data exceeds the shutdown alarm threshold, the corresponding equipment is shut down, the fault is located, and then the fault is reconstructed based on the fault location results.

[0012] As a preferred embodiment of the present invention, when making a steady-state judgment, the historical test data of the monitoring data are analyzed based on the test data to see if they conform to a Gaussian distribution. If they do, a Gaussian distribution model is used for steady-state judgment; if they do not, a steady-state detection method based on non-assumed data distribution is used for steady-state judgment.

[0013] As a preferred embodiment of the present invention, when using a steady-state detection method based on non-assumed data distribution for steady-state judgment, a classification algorithm is used to obtain a single-parameter steady-state detection model for the monitoring data. At the end of each voyage, the false alarms and missed alarms during the voyage are evaluated, and the parameters in the steady-state detection model are updated.

[0014] In a preferred embodiment of the present invention, during the trend judgment process, if an abnormal trend is determined, the difference between the predicted data of the current monitoring data and the set shutdown alarm threshold is further judged. If the difference is greater than the set value, fault location is performed first; otherwise, it indicates that there is a risk of failure soon according to the current trend, and the corresponding equipment is shut down while fault location is performed.

[0015] In a preferred embodiment of the present invention, when locating a fault, the fault is located based on a preset expert knowledge base; the expert knowledge base stores the fault types that cause abnormalities in each monitoring data and the probability that each fault type causes abnormalities in the monitoring data; the fault is located by matching the fault type with the highest probability of causing the abnormality in the monitoring data based on the expert knowledge base.

[0016] As a preferred embodiment of the present invention, sensing devices are used to collect operating data of various devices in a comprehensive power system from the perspectives of thermal, electrical, and vibration, thereby obtaining monitoring data.

[0017] In a preferred embodiment of the present invention, when performing fault reconstruction, it is first determined whether the device corresponding to the monitoring data has the function of reduced operating conditions. If it has the function of reduced operating conditions, the corresponding device is subjected to reduced operating conditions processing; if it does not have the function of reduced operating conditions, the corresponding device is subjected to shutdown processing.

[0018] As a preferred embodiment of the present invention, after the faulty equipment is degraded or shut down, it is further determined whether the integrated power system can still meet the needs of normal operation. If it can, it will operate normally under the current conditions; otherwise, the backup equipment will be activated.

[0019] Beneficial effects:

[0020] (1) The health management method of the unmanned ship integrated power system of the present invention adopts early warning based on fault prediction, thereby realizing fault early warning / pre-location and fault reconstruction / pre-reconstruction of the unmanned ship integrated power system, ensuring the healthy operation of its integrated power system during the operation of the unmanned ship.

[0021] (2) The health management method of the unmanned vessel integrated power system of the present invention adopts a hierarchical alarm mechanism, which can improve the availability of the integrated power system.

[0022] (3) Given that there is a certain proportion of identification “errors” in the steady-state judgment of data when the equipment is just started or when the operating conditions are switched, the health management method of the unmanned ship integrated power system of the present invention further adds trend judgment after steady-state judgment. Through trend analysis, some of the “errors” in steady-state judgment can be suppressed, and the accuracy of fault judgment can be improved.

[0023] (4) When using a steady-state detection method based on non-assumed data distribution to make steady-state judgment, a steady-state detection model of the monitoring data can be obtained. At the end of each voyage, experts evaluate the false alarms and missed alarms during the voyage and update the parameters in the steady-state detection model, thereby enhancing the robustness of the steady-state detection model.

[0024] (5) During the trend judgment process, after judging the trend abnormality, further judge the difference between the predicted data of the current monitoring data and the set shutdown alarm threshold. If the difference between the two is large, enter the fault location module to locate the fault; otherwise, it indicates that there is a risk of fault soon according to the current trend, and implement shutdown operation for the corresponding equipment, and start fault location at the same time; thus, timely fault warning can be guaranteed and safety can be ensured.

[0025] (6) The present invention directly uses a preset expert knowledge base to locate faults, which is simple and reliable. Attached Figure Description

[0026] Figure 1 This is a flowchart of the unmanned vessel integrated power system health management method of the present invention. Detailed Implementation

[0027] The present invention will now be described in further detail with reference to the accompanying drawings and embodiments.

[0028] This embodiment provides a health management method for an unmanned vessel integrated power system. This health management method is based on an autonomous decision-making model, which can achieve pre-reconfiguration, improve the availability of the vessel's integrated power system, and ensure the healthy operation of the unmanned vessel's integrated power system.

[0029] The health management method for the integrated power system of the unmanned vessel includes the following steps: data acquisition and processing, alarm judgment, steady state judgment, trend judgment, fault location and fault reconstruction.

[0030] Steady-state and trend judgments are primarily data-driven, and the algorithms are updated and optimized adaptively based on real-time data through learning algorithms; fault location adopts a combination of data-driven and expert knowledge bases; fault reconstruction is mainly constructed through expert knowledge bases.

[0031] The unmanned surface vessel's integrated power system includes: a diesel generator set, a propulsion motor, a lithium battery pack, a cooling pump, a switchboard, and corresponding auxiliary equipment. During data acquisition and processing, the data acquisition and processing module utilizes sensors to obtain data reflecting the operational information of the aforementioned equipment within the integrated power system. The module primarily collects operational data from multiple perspectives, including thermal (but not limited to temperature, flow rate, and pressure), electrical (but not limited to voltage and current), and vibration (but not limited to the vibration of the diesel generator set and cooling pump). The collected data includes, but is not limited to: diesel engine exhaust temperature, generator stator winding temperature, generator set cooling freshwater temperature, propulsion motor speed, lithium battery pack temperature, diesel generator set vibration, and cooling pump vibration.

[0032] The data acquisition and processing module sends the acquired data to the alarm judgment module in real time for alarm judgment. The alarm judgment module has two preset alarm thresholds for each monitoring data: a shutdown alarm threshold and a reduced operating condition alarm threshold.

[0033] When the alarm judgment module determines that all monitoring data are less than the corresponding de-operation alarm threshold, it indicates that the current integrated power system is operating normally and outputs a normal operation command.

[0034] When the alarm judgment module determines that a certain monitoring data is greater than the corresponding de-operation alarm threshold but less than the corresponding shutdown alarm threshold, it sends the monitoring data to the steady-state judgment module for steady-state judgment.

[0035] When the alarm judgment module determines that a certain monitoring data exceeds the shutdown alarm threshold, it first shuts down the corresponding equipment, performs fault location, and then reconstructs the fault based on the fault location results.

[0036] The steady-state judgment module performs a steady-state assessment on the monitoring data sent by the alarm judgment module. If the monitoring data is determined to be fluctuating within a small range, it is considered to be in a steady state, indicating that the current integrated power system is operating normally, and a normal operation command is output. Otherwise, the monitoring data is considered to be in a non-steady state, and the monitoring data is further sent to the trend judgment module for trend analysis. The reason for further trend analysis is that when the equipment is first started up or when switching operating conditions, there is a certain proportion of identification "errors" in the steady-state judgment of the data. For monitoring data whose steady-state judgment result is "non-steady state", further trend analysis is needed to determine whether there is a real anomaly. Trend analysis can suppress some of the "errors" in steady-state judgment.

[0037] When making steady-state judgments: First, a pre-analysis based on experimental data is performed, that is, the probability distribution model of the monitoring data is selected based on the distribution of historical data, and the experimental data of the monitoring data is tested by a distribution test method (such as the Kolmogorov-Smirnov test method) to see if the experimental data of the monitoring data conforms to a Gaussian distribution. If it does, the Gaussian distribution model is adopted; if it does not, a steady-state detection method based on non-assumed data distribution is adopted.

[0038] When using a Gaussian distribution model, if the monitored data is within the Gaussian distribution range (i.e., concentrated around a certain data point), it is considered to be in a steady state; otherwise, it is considered to be in a non-steady state.

[0039] When using a steady-state detection method based on non-assumed data distribution, a classification algorithm is used to determine whether there are outliers in the monitoring data. If they exist, it indicates that the monitoring data is in a non-steady state; if they do not exist, it indicates that the monitoring data is in a steady state.

[0040] When using a steady-state detection method based on non-assumed data distribution for steady-state judgment, a steady-state detection model for the monitoring data can be obtained. Regarding the real-time updating and reusability of the steady-state detection model, considering the high cost of real-time model updating at the ship's end during navigation and the tendency of optimization anomalies to favor the data-driven approach, at the end of each voyage, experts evaluate the false alarms and missed alarms during the voyage. Based on this, the optimization process is calibrated in reverse, and the parameters in the steady-state detection model are updated, thereby enhancing the robustness of the steady-state detection model.

[0041] The trend judgment module further analyzes the monitoring data identified as "non-steady" by the steady-state judgment module to predict its trend within a set time range. This trend judgment utilizes historical data and a predictive model to calculate future data. If the current monitoring data shows a normal trend within the set time range, it indicates that the integrated power system is operating normally, and a normal operation command is output. If the current monitoring data shows an abnormal trend within the set time range, the module further judges the difference between the predicted data and the set shutdown alarm threshold. If this difference is greater than the set value, indicating a significant gap, the module initiates fault location to locate the fault. Otherwise, it indicates a high risk of failure based on the current trend, and a shutdown operation is performed on the corresponding equipment, while fault location is activated.

[0042] Trend analysis can enable fault prediction and early warning, thereby achieving pre-reconfiguration.

[0043] When making data predictions, the trend judgment module uses the monitoring data before the current time as input and uses the selected prediction model (such as a regression model) to predict the data within a set time range in the future. It then analyzes the changing trend of the current monitoring data to obtain the predicted data at the set time point in the future. The predicted data is then compared with the experimental data at the set time point. If the difference is large (i.e. greater than the preset value), it indicates that the trend of the current monitoring data is abnormal; otherwise, it is considered that the trend of the current monitoring data is normal.

[0044] Fault localization employs a combination of data-driven approaches and an expert knowledge base. The fault localization module has a pre-installed expert knowledge base containing the causes (fault types) leading to anomalies in various monitoring data, along with the probability of each cause causing the anomaly (typically, multiple causes exist for a given monitoring data anomaly, each with a different probability). When fault localization is required, the module matches the fault type with the highest probability of causing the anomaly in that monitoring data based on the expert knowledge base, thereby locating the fault.

[0045] Based on the fault location results, the fault reconfiguration module generates a reconfiguration strategy and outputs reconfiguration instructions to the corresponding devices to perform fault reconfiguration or pre-reconfiguration (reconfiguration refers to reconfiguration under shutdown conditions, and pre-reconfiguration refers to reconfiguration under abnormal trend conditions and no shutdown conditions as determined by the trend judgment module).

[0046] When performing fault reconstruction, the fault reconstruction module first determines whether the device corresponding to the monitored data has a de-operating condition function. If it does, it performs de-operating condition processing on the corresponding device; otherwise, it outputs a shutdown command. Simultaneously, the fault reconstruction module further determines whether the integrated power system can still meet normal operation requirements after shutting down or de-operating the faulty device, and decides whether to activate backup equipment. If the integrated power system meets normal operation requirements after shutting down or de-operating the faulty device, it will operate normally under the current conditions; otherwise, the backup equipment for the faulty device will be activated.

[0047] By downgrading the operating conditions of faulty equipment or activating the backup equipment of faulty equipment, the reconfiguration / pre-reconfiguration of the unmanned vessel's integrated power system can be achieved. After the reconfiguration / pre-reconfiguration is completed, health management of the reconfigured / pre-reconfigured unmanned vessel's integrated power system continues.

[0048] Although the present invention has been described in detail above with general descriptions and specific embodiments, modifications or improvements can be made to it, which will be obvious to those skilled in the art. Therefore, all such modifications or improvements made without departing from the spirit of the present invention fall within the scope of protection claimed by the present invention.

Claims

1. A method for health management of an unmanned surface vessel's integrated power system, characterized by: During the operation of the unmanned vessel, real-time monitoring can reflect the operating information of various equipment in the integrated power system; The system also performs alarm judgments on each monitoring data point: A tiered alarm approach is used, with two preset alarm thresholds for each monitoring data point: a shutdown alarm threshold and a reduced operating condition alarm threshold. When all monitoring data are less than the corresponding de-operation alarm threshold, it indicates that the current integrated power system is operating normally; When a monitoring data point is greater than the corresponding de-operation alarm threshold but less than the corresponding shutdown alarm threshold, a steady-state judgment is performed on the monitoring data. If the monitoring data is determined to be fluctuating only within the set range, it is considered to be in a steady state, indicating that the current integrated power system is operating normally. Otherwise, the monitoring data is considered to be in a non-steady state, and the monitoring data is further sent to the trend judgment module for trend judgment. The trend judgment can obtain the predicted data of the current monitoring data within a set time range, and the predicted data can be used to determine whether the current monitoring data is a normal trend or an abnormal trend within the set time range. If the trend is normal, it indicates that the current integrated power system is operating normally. If the trend is abnormal, fault location is performed, and then fault reconstruction is performed based on the fault location results to achieve pre-reconstruction. When a certain monitoring data exceeds the shutdown alarm threshold, the corresponding equipment is shut down, the fault is located, and then the fault is reconstructed based on the fault location results. When making a steady-state judgment, the historical test data of the monitoring data are analyzed based on the test data to see if they conform to a Gaussian distribution. If they do, the Gaussian distribution model is used for steady-state judgment; if they do not, a steady-state detection method based on non-assumed data distribution is used for steady-state judgment. When using a steady-state detection method based on non-assumed data distribution for steady-state judgment, a classification algorithm is used to obtain a single-parameter steady-state detection model for the monitoring data. At the end of each voyage, the false alarms and missed alarms during the voyage are evaluated, and the parameters in the steady-state detection model are updated. During the trend judgment process, if an abnormal trend is determined, the difference between the predicted data of the current monitoring data and the set shutdown alarm threshold is further judged. If the difference is greater than the set value, the fault location is performed first; otherwise, it indicates that there is a risk of failure soon according to the current trend, and the corresponding equipment is shut down and the fault location is performed at the same time.

2. The health management method for the integrated power system of an unmanned vessel as described in claim 1, characterized in that: When locating a fault, the fault is located based on a preset expert knowledge base. The expert knowledge base contains the fault types that cause abnormalities in each monitoring data and the probability that each fault type causes abnormalities in the monitoring data. The fault is located by matching the fault type with the highest probability of causing the abnormality in the monitoring data based on the expert knowledge base.

3. The health management method for the integrated power system of an unmanned vessel as described in claim 1, characterized in that: By using sensing devices to collect operational data of various devices in a comprehensive power system from the perspectives of thermal, electrical, and vibration, monitoring data can be obtained.

4. The health management method for the integrated power system of an unmanned vessel as described in claim 1, characterized in that: When refactoring a fault, first determine whether the device corresponding to the monitoring data has the function of reduced operating conditions. If it has the function of reduced operating conditions, then the corresponding device is processed to reduce operating conditions; if it does not have the function of reduced operating conditions, then the corresponding device is shut down.

5. The health management method for the integrated power system of an unmanned vessel as described in claim 4, characterized in that: After reducing the operating conditions or shutting down the faulty equipment, it is further determined whether the integrated power system can still meet the needs of normal operation. If it can, it will operate normally under the current conditions; otherwise, the backup equipment will be activated.

Images