CN116136926A - Underwater robot health management system based on Bayesian network - Google Patents
Underwater robot health management system based on Bayesian network Download PDFInfo
- Publication number
- CN116136926A CN116136926A CN202111367676.0A CN202111367676A CN116136926A CN 116136926 A CN116136926 A CN 116136926A CN 202111367676 A CN202111367676 A CN 202111367676A CN 116136926 A CN116136926 A CN 116136926A
- Authority
- CN
- China
- Prior art keywords
- health
- life
- underwater robot
- node
- index
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/10—Geometric CAD
- G06F30/17—Mechanical parametric or variational design
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2119/00—Details relating to the type or aim of the analysis or the optimisation
- G06F2119/04—Ageing analysis or optimisation against ageing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Geometry (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- Computer Hardware Design (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Analysis (AREA)
- Pure & Applied Mathematics (AREA)
- Mathematical Optimization (AREA)
- Computational Mathematics (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention relates to the field of fault diagnosis and health management, in particular to an underwater robot health management system based on a Bayesian network. Comprising the following steps: the health evaluation sub-module is used for acquiring the Bayesian information file in the database and the underwater robot node information, constructing a Bayesian network according to the Bayesian information file, inputting the underwater robot node information into the Bayesian network to obtain the health degree probability of the node, and further obtaining the health index for describing the health condition of the node. The failure prediction sub-module is used for obtaining the life information file of the underwater robot node, constructing a life model, taking the life information file of the underwater robot node as the input of the life model, and obtaining a life index for describing the aging degree of the node. The invention does not need to establish a mathematical model with accurate system, avoids the influence of modeling errors and related uncertainties, and can fully utilize the data in the database to better meet the actual requirements of engineering.
Description
Technical Field
The invention relates to the field of fault diagnosis and health management, in particular to an underwater robot health management system based on a Bayesian network.
Background
The fault diagnosis is a new subject developed in recent 40 years, and mainly uses various detection methods to judge the running state and abnormal condition of the system and diagnose the fault type, position and fault cause. The method is an edge discipline of multi-science comprehensive application formed to adapt to engineering actual demands, and the theoretical basis is modern control theory, computer technology, mathematical statistics, signal processing, pattern recognition, artificial intelligence, artificial neural network and other corresponding application disciplines.
In practical application fields such as deep sea salvage, space exploration, ruin rescue, battlefield projectile removal, underground mining, nuclear waste removal, polar scientific investigation and the like, a worker wants to personally visit a site to take a lot of financial resources and material resources, or no condition for human access or operation at all. At this time, the robot is undoubtedly not the second choice to perform dangerous tasks instead of humans. The robot can monitor and control own behaviors in real time, and can automatically process in time when a system fails. That is, if a malfunction of a robot can be detected and predicted in time and the robot in the malfunction can be restored to a normal operation state, the performance of such a robot must be greatly improved.
At present, the development of machine learning and other technologies is applied to practical engineering, so that the intelligent degree of a fault diagnosis method is greatly improved. In view of its extremely strong practicality, intelligent diagnosis technology has become a research hotspot in the field of fault diagnosis today on a global scale.
The existing mature health management method based on data driving in other fields is not fully used for mining multi-source data and similar sample information in the practical application of the underwater robot control field. The existing research results show that: on the one hand, alternating ambient temperature is an important factor causing performance degradation of key components of the underwater robot system; the telemetering data can truly reflect the real-time running state and the performance change process of the system. On the other hand, data such as development, test and experiment of the control system and life information of the same type of system and retirement can provide important information support for researching the residual life of the current system. The existing method does not fully mine the effective information implicit in the multi-source data, and the service life of the statistical control system is extremely short, so that the existing method is difficult to be directly applied to a satellite control system. The combination of data in the health management system of the underwater robot becomes a research focus. The health management system is combined with the real-time data to realize quantitative analysis and expression of different health degrees, and the health management system is expressed in a score form in a health evaluation mode.
Disclosure of Invention
The invention aims to provide a health management system based on an underwater robot, which overcomes the defects.
The technical scheme adopted by the invention for achieving the purpose is as follows:
an underwater robot health management system based on a bayesian network, comprising:
the health evaluation sub-module is used for acquiring the Bayesian information file in the database and the underwater robot node information, constructing a Bayesian network according to the Bayesian information file, inputting the underwater robot node information into the Bayesian network to obtain the health degree probability of the node, and further obtaining the health index for describing the health condition of the node.
The failure prediction sub-module is used for obtaining the life information file of the underwater robot node, constructing a life model, taking the life information file of the underwater robot node as the input of the life model, and obtaining a life index for describing the aging degree of the node.
Each component of the underwater robot is used as a node in the Bayesian network, and the Bayesian information file comprises reliability information and influence information of the node.
The node information comprises which sub-nodes are contained in the node, and the weight, the updating state and scoring method, the using time and the using times of the sub-nodes.
The underwater robot node life information file comprises: life decay function parameters, life decay function correlation characteristics.
The lifetime model is a lifetime decay function L,
wherein a is 0 、、λ、a 1 Are all constant, lambda and a 0 Defining the decay speed of life index, a 1 Define the initial decay state of life index, L start Is the initial value of lifetime.
After the health index and the life index of a certain part of the underwater robot are obtained, the whole health index and the life index of the underwater robot are obtained by using a step-by-step detection method.
The step-by-step detection method specifically comprises the following steps:
respectively taking a health index and a life index of a certain part of the underwater robot as inputs of a health evaluation sub-module and a fault prediction sub-module to obtain the health index and the life index of a subsystem where the part is positioned;
and respectively inputting the health index and the life index of the subsystem as the inputs of the health evaluation sub-module and the fault prediction sub-module to obtain the health index and the life index of the underwater robot.
The underwater robot health management method based on the Bayesian network comprises the following steps:
the health evaluation submodule acquires the Bayesian information file and the underwater robot node information in the database, constructs a Bayesian network according to the Bayesian information file, and inputs the underwater robot node information into the Bayesian network to obtain the health index of the node.
The failure prediction submodule acquires a life information file of the underwater robot node, a life model is built, the life information file of the underwater robot node is used as input of the life model, and a life index of the node is obtained.
A computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements the bayesian network based underwater robot health management method.
The invention has the following beneficial effects and advantages:
the invention uses Bayesian network to take probabilistic reasoning as the main principle, has solid theoretical basis, and has mature inference algorithm and accurate inference result of naive Bayesian, and has lower prior probability requirement. The system accurate mathematical model is not required to be established, the influence of modeling errors and related uncertainties is avoided, and the data in the database can be fully utilized to better meet the actual requirements of engineering.
Drawings
FIG. 1 is a block diagram of a health assessment of the present invention;
FIG. 2 is a block diagram of a fault prediction module of the present invention;
FIG. 3 is a system configuration diagram of the present invention;
FIG. 4 is a schematic diagram of a health management module hierarchy of the present invention;
FIG. 5 is a schematic diagram of the functionality and output of a node computer according to the present invention;
FIG. 6 is a schematic diagram of the functions and outputs of a management decision computer according to the present invention;
fig. 7 is a logic flow diagram of the present invention.
Detailed Description
The present invention will be described in further detail with reference to the accompanying drawings and examples.
An underwater robot-based health management system, the system comprising two modules, health assessment and fault prediction, comprising:
(1) And reading the parameter file by a file reading sub-module to obtain parameters for constructing the Bayesian network. And constructing a Bayesian network by the health evaluation submodule according to the Bayesian network parameters. The reliability information and the node influence information of the nodes are read from the file, each component is one node in the Bayesian network, the influence of the node on other nodes is realized through node modes, and each mode has a corresponding reliability degree value as a probability value in the Bayesian network.
(2) The health evaluation submodule reads the modes of the components and infers the health states of the components according to the modes. Establishing an inference engine for the formed Bayesian network to perform inference through the obtained information;
(3) The health evaluation sub-module continuously acquires the mode information of the component, transmits the mode information as input to the Bayesian network, obtains the current mode of the component subsystem through reasoning, obtains the health index of the component through calculation, and outputs the health index and the health information of the component;
(4) The fault prediction submodule firstly reads the life information file, acquires parameters required by a life model and establishes a part life model;
(5) The change in component pattern is input information for failure prediction. The failure prediction sub-module calculates a life index of the component from the component modes; and calculating the life index of the subsystem according to the life weighting parameter and the part life index by the life model of the subsystem.
The system health management software is generally designed to:
the health management system evaluates the health state of the system by inputting the state information of the components and the subsystems, and predicts the faults of the components, the subsystems and the systems according to the life model and the fault information.
The system comprises two modules, namely health assessment and fault prediction. The health assessment module monitors the health state of each level of system/subsystem according to the data information, and realizes each level of health assessment through construction of an inference network and a weighting function. And carrying out fault prediction on the system/subsystem according to the life model and the extraction of fault characteristics, and predicting the future aging trend of the system.
The health evaluation and fault prediction sub-module respectively operates in the secondary node computer and the management decision computer, and evaluates and predicts the faults of the components and the subsystems respectively.
Optionally the system secondary node computer module is designed to:
for node computers, the health management system functions as a component health assessment and life model calculation. The obtained component-level health information includes: 1. a component level health indicator; 2. component level health information; 3. a component level lifetime index; 4. component level lifetime information.
The health management module needs to input data related to the component for health assessment. The fault evaluation sub-module obtains the health index of the component through the established Bayesian network and the component fault mode change reasoning. The failure prediction sub-module calculates a life index value and life information according to the current failure mode of the component through the established life model.
The health assessment and fault prediction sub-module is directly component oriented at the node computer level. The sub-module directly processes the information of the component level and outputs the result, on one hand, the node computer directly processes the input information of the component, and the health evaluation and fault prediction result of the component is obtained through the health evaluation and fault prediction module, on the other hand, the component level result becomes the input information of the subsystem level health evaluation and fault prediction in the management decision computer. Because the subsystem is made up of components, the health and life conditions of the subsystem are weighted by the components.
The system health evaluation module method comprises the following steps:
naive bayes, the formula is:
wherein, P (B) and the conditional probability P (A|B) are both considered to be given by an expert system, and P (A) can be obtained by a full probability formula.
Life decay function:
wherein a is 0 、、λ、a 1 Are all constant, lambda and a 0 Defining the decay speed of life index, a 1 Define the initial decay state of life index, L start For the initial lifeValues.
FIG. 1 is a block diagram of a health assessment of the present invention.
(1) The health evaluation submodule reads the modes of the components and infers the health states of the components according to the modes. Establishing an inference engine for the formed Bayesian network to perform inference through the obtained information;
(2) The health evaluation sub-module continuously acquires the mode information of the component, transmits the mode information as input to the Bayesian network, obtains the current mode of the component subsystem through reasoning, obtains the health index of the component through calculation, and outputs the health index and the health information of the component;
fig. 2 is a block diagram of the failure prediction of the present invention.
(1) The fault prediction submodule firstly reads the life information file, acquires parameters required by a life model and establishes a part life model;
(2) The change in component pattern is input information for failure prediction. The failure prediction sub-module calculates a life index of the component from the component modes; and calculating the life index of the subsystem according to the life weighting parameter and the part life index by the life model of the subsystem.
Fig. 3 is a system configuration diagram of the present invention.
The invention relates to an underwater robot based health management system, wherein required input information comprises preset parameters and information input information, and output information is obtained. The parameters are stored in a mode of a jason file and read by a preprocessing program in the system; the health assessment module monitors the health states of all levels of systems/subsystems according to the data information by using a Bayesian network and a life decay function method, and achieves all levels of health assessment through construction of an inference network and a weighting function. The fault prediction module predicts the faults of the system/subsystem according to the life model and the extraction of the fault characteristics, and predicts the future aging trend of the system.
Fig. 4 is a schematic diagram of a health management module hierarchy of the present invention.
The health assessment and fault prediction sub-module is directly component oriented at the node computer level. The sub-module directly processes the information of the component level and outputs the result, on one hand, the node computer directly processes the input information of the component, and the health evaluation and fault prediction result of the component is obtained through the state evaluation and fault prediction module, on the other hand, the component level result becomes the input information of the subsystem level health evaluation and fault prediction in the management decision computer. Because the subsystem is made up of components, the health and life conditions of the subsystem are weighted by the components.
Fig. 5 is a schematic diagram of the functions and output of the node computer according to the present invention.
For node computers, the health management module functions as bayesian network reasoning and life model calculation. The obtained component-level health information includes: 1. a component level health indicator; 2. component level health information; 3. a component level lifetime index; 4. component level lifetime information.
FIG. 6 is a schematic diagram of the functions and outputs of the management decision computer of the present invention.
The health management module weights the health indexes of the components in the subsystem according to the weights of the components in the subsystem in the function of the management decision computer according to the health indexes, the health information and the key parameters of the component level to obtain the health indexes of the subsystem; the health management module obtains subsystem-level life indexes through analysis weighting of the component-level life information. Meanwhile, the influence of the key information on the life index is weighted and considered, and an output result of the subsystem is obtained.
FIG. 7 is a flow chart of the logic of the present invention
The method comprises the following steps:
(1) And reading the parameter file by a file reading sub-module to obtain parameters for constructing the Bayesian network. And constructing a Bayesian network by the health evaluation submodule according to the Bayesian network parameters. The reliability information and the node influence information of the nodes are read from the file, each component is one node in the Bayesian network, the influence of the node on other nodes is realized through node modes, and each mode has a corresponding reliability degree value as a probability value in the Bayesian network.
(2) The health evaluation submodule reads the modes of the components and infers the health states of the components according to the modes. Establishing an inference engine for the formed Bayesian network to perform inference through the obtained information;
(3) The health evaluation sub-module continuously acquires the mode information of the component, transmits the mode information as input to the Bayesian network, obtains the current mode of the component subsystem through reasoning, obtains the health index of the component through calculation, and outputs the health index and the health information of the component;
(4) The fault prediction submodule firstly reads the life information file, acquires parameters required by a life model and establishes a part life model;
(5) The change in component pattern is input information for failure prediction. The failure prediction sub-module calculates a life index of the component from the component modes; and calculating the life index of the subsystem according to the life weighting parameter and the part life index by the life model of the subsystem.
Claims (9)
1. The utility model provides an underwater robot health management system based on Bayesian network which characterized in that includes:
the health evaluation sub-module is used for acquiring the Bayesian information file in the database and the underwater robot node information, constructing a Bayesian network according to the Bayesian information file, inputting the underwater robot node information into the Bayesian network to obtain the health degree probability of the node, and further obtaining the health index for describing the health condition of the node;
the failure prediction sub-module is used for obtaining the life information file of the underwater robot node, constructing a life model, taking the life information file of the underwater robot node as the input of the life model, and obtaining a life index for describing the aging degree of the node.
2. The bayesian network based underwater robot health management system of claim 1, wherein each component of the underwater robot is a node in the bayesian network, and wherein the bayesian information file comprises reliability information and influence information of the node.
3. The bayesian network based underwater robot health management system of claim 1, wherein the node information comprises which sub-nodes are included in the node, and the weights, update status and scoring methods, use times, and number of uses of the sub-nodes.
4. The bayesian network-based underwater robot health management system of claim 1, wherein the underwater robot node lifetime information file comprises: life decay function parameters, life decay function correlation characteristics.
5. The system of claim 1, wherein the life model is a life decay function L,
wherein a is 0 、λ、a 1 Are all constant, lambda and a 0 Defining the decay speed of life index, a 1 Define the initial decay state of life index, L start Is the initial value of lifetime.
6. The system for managing the health of an underwater robot based on a bayesian network according to claim 1, wherein the health index and the life index of a certain part of the underwater robot are obtained, and then the health index and the life index of the whole underwater robot are obtained by using a step-by-step detection method.
7. The underwater robot health management system based on the bayesian network according to claim 6, wherein the step-by-step detection method specifically comprises:
respectively taking a health index and a life index of a certain part of the underwater robot as inputs of a health evaluation sub-module and a fault prediction sub-module to obtain the health index and the life index of a subsystem where the part is positioned;
and respectively inputting the health index and the life index of the subsystem as the inputs of the health evaluation sub-module and the fault prediction sub-module to obtain the health index and the life index of the underwater robot.
8. The underwater robot health management method based on the Bayesian network is characterized by comprising the following steps of:
the health evaluation submodule acquires a Bayesian information file and underwater robot node information in a database, constructs a Bayesian network according to the Bayesian information file, and inputs the underwater robot node information into the Bayesian network to obtain the health index of the node;
the failure prediction submodule acquires a life information file of the underwater robot node, a life model is built, the life information file of the underwater robot node is used as input of the life model, and a life index of the node is obtained.
9. A computer readable storage medium, wherein a computer program is stored on the storage medium, which when executed by a processor, implements the bayesian network based underwater robot health management method of claim 8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111367676.0A CN116136926A (en) | 2021-11-18 | 2021-11-18 | Underwater robot health management system based on Bayesian network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202111367676.0A CN116136926A (en) | 2021-11-18 | 2021-11-18 | Underwater robot health management system based on Bayesian network |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN116136926A true CN116136926A (en) | 2023-05-19 |
Family
ID=86334179
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202111367676.0A Pending CN116136926A (en) | 2021-11-18 | 2021-11-18 | Underwater robot health management system based on Bayesian network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116136926A (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116910920A (en) * | 2023-09-12 | 2023-10-20 | 陕西万禾数字科技有限公司 | Aeroengine comprehensive health management system and method based on augmented reality technology |
-
2021
- 2021-11-18 CN CN202111367676.0A patent/CN116136926A/en active Pending
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116910920A (en) * | 2023-09-12 | 2023-10-20 | 陕西万禾数字科技有限公司 | Aeroengine comprehensive health management system and method based on augmented reality technology |
| CN116910920B (en) * | 2023-09-12 | 2023-12-05 | 陕西万禾数字科技有限公司 | Aeroengine comprehensive health management system and method based on augmented reality technology |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Lin et al. | Development and assessment of a nearly autonomous management and control system for advanced reactors | |
| Elattar et al. | Prognostics: a literature review | |
| CN110807257A (en) | Method for predicting residual life of aircraft engine | |
| Phillips et al. | A knowledge driven approach to aerospace condition monitoring | |
| JP2008537262A (en) | Decision support method and system | |
| US20230017065A1 (en) | Computing an explainable event horizon estimate | |
| KR102656115B1 (en) | Remaining useful life prediction method of fuel cell system and digital twin device performing the same | |
| Choo et al. | Adaptive multi-scale prognostics and health management for smart manufacturing systems | |
| Siraskar et al. | Reinforcement learning for predictive maintenance: A systematic technical review | |
| CN116819336A (en) | Power battery abnormality detection system based on unsupervised learning | |
| Li et al. | Intelligent reliability and maintainability of energy infrastructure assets | |
| CN116136926A (en) | Underwater robot health management system based on Bayesian network | |
| Peng et al. | Health indicator construction based on multisensors for intelligent remaining useful life prediction: A reinforcement learning approach | |
| Liu et al. | Deep learning approach for sensor data prediction and sensor fault diagnosis in wind turbine blade | |
| CN117689373A (en) | Maintenance decision support method for energy router of flexible direct-current traction power supply system | |
| Zhang et al. | Hierarchical hybrid testability modeling and evaluation method based on information fusion | |
| CN116611523B (en) | Method and system for predicting interpretable faults of turbofan engine | |
| Song et al. | Research on preventive maintenance strategies and systems for in-service ship equipment | |
| Kurtoglu et al. | A framework for systematic benchmarking of monitoring and diagnostic systems | |
| CN116384087A (en) | Manned deep diving operation human factor reliability analysis method, system and equipment | |
| Borth et al. | Probabilistic health and mission readiness assessment at system-level | |
| KR20230122269A (en) | Industrial facility diagnosis prediction system using AI technology | |
| Lin et al. | Enhancing the operational resilience of advanced reactors with digital twins by recurrent neural networks | |
| Zhang et al. | Product quality reliability analysis based on rough Bayesian network | |
| Natsumeda et al. | RULENet: end-to-end learning with the dual-estimator for remaining useful life estimation |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| SE01 | Entry into force of request for substantive examination | ||
| SE01 | Entry into force of request for substantive examination |