New energy vehicle roll-on and roll-off transportation risk assessment method and device
By constructing a Bayesian network and combining the confidence index and weights of expert judgments, the problem of multivariate data processing in the Roll-Round Transportation Safety Assessment of New Energy Vehicles is solved, and the reliability and accuracy of systemic risk assessment is achieved.
Patent Information
- Application Number
- CN202510427103.4
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2025-04-07
- Publication Date
- 2025-08-01
AI Technical Summary
The existing Roll-Road Transport Safety Assessment Technology of new energy vehicles lacks effective means of processing multiple data, and it is difficult to meet the needs of systematic safety analysis in complex variable environments.
Build a Bayesian network, combine the confidence index and weights judged by experts, and reason the risk level of roll-on and roll-on transportation of new energy vehicles through Bayesian network, integrate evaluation indicators and conduct systematic analysis using Bayesian network model.
It improves the reliability and accuracy of roll-on and roll-in transportation risk assessment of new energy vehicles, reduces the impact of experts' subjective judgments, and provides a systematic risk assessment method.
Smart Images
Figure CN120410187A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of roll-on / roll-off transportation technology, and in particular to a roll-on / roll-off transportation risk assessment method and device for new energy vehicles. Background Art
[0002] With the rapid development of my country's new energy vehicle industry, the number of domestic new energy vehicles continues to grow, and the scale of new energy vehicle transportation on ro-ro vessels is increasing. However, new energy vehicle fires are frequent, and thermal runaway of power batteries and the resulting spontaneous combustion have become a focus of industry attention. Risks associated with ro-ro transportation of new energy vehicles are compounded by the difficulties of rescue and escape at sea, the high risk of fire fighting, and the potential for damage to the ship's structure due to fire spread. Therefore, it is urgent to strengthen research on the identification and risk assessment of safety hazards in ro-ro transportation of new energy vehicles, and to improve fire risk prevention in ro-ro transportation of new energy vehicles.
[0003] The existing safety assessment technology for roll-on / roll-off transportation of new energy vehicles lacks effective means to process multivariate data and is unable to meet the needs of systematic safety analysis in a complex variable environment. Summary of the Invention
[0004] In view of this, it is necessary to provide a new energy vehicle roll-on / roll-off transportation risk assessment method and device to solve the problem that the existing new energy vehicle roll-on / roll-off transportation safety assessment technology lacks effective means to process multivariate data and is difficult to meet the needs of systematic safety analysis in a complex variable environment.
[0005] In order to solve the above problems, in a first aspect, the present invention provides a method for assessing the risk of roll-on / roll-off transportation of new energy vehicles, comprising: Constructing a Bayesian network based on the relationship between the evaluation indicators of the roll-on / roll-off transportation risk of new energy vehicles and the relationship between the evaluation indicators and the risk level; Obtaining a first judgment result of an expert on the occurrence probability of each node state of the root node in the Bayesian network; the first judgment result includes: a target probability level and a confidence index for the target probability level; The product of the expert weight and the confidence index is used as the predicted probability of the target probability level; The roll-on / roll-off transportation risk level of new energy vehicles is obtained based on the Bayesian network, the target probability level and the predicted probability reasoning.
[0006] In a possible implementation, the new energy vehicle roll-on / roll-off transportation risk level inferred based on the Bayesian network, the target probability level, and the predicted probability includes: determining the predicted probability that the occurrence probability of the node state is at other probability levels according to the probability distribution characteristics and the predicted probability of the target probability level; wherein, the probability distribution characteristic is a normal distribution characteristic with the predicted probability of the target probability level as the center and the predicted probabilities of the probability levels on both sides decreasing; inferring the new energy vehicle roll-on / roll-off transportation risk level based on the Bayesian network and the predicted probabilities of each probability level.
[0007] In a possible implementation, there are multiple experts, and there are 5 probability levels; the predicted probability is determined by the following formula:
[0008]
[0009]
[0010] Wherein, represents the predicted probability that the occurrence probability of the node state is at the probability level i , represents the predicted probability that the expert n judges that the node state is at the target probability level k , represents the lower bound of the first probability level, represents the upper bound of the first probability level and the lower bound of the second probability level, represents the upper bound of the second probability level and the lower bound of the third probability level, represents the upper bound of the third probability level and the lower bound of the fourth probability level, represents the upper bound of the fourth probability level and the lower bound of the fifth probability level, represents the upper bound of the fifth probability level.
[0011] In a possible implementation, the inferring the new energy vehicle roll-on / roll-off transportation risk level based on the Bayesian network and the predicted probabilities of each probability level includes: synthesizing the predicted probabilities of each probability level of the node state to generate the target predicted probability of each probability level; calculating the triangular fuzzy number of the node state according to the target predicted probabilities of each probability level of the node state, and then performing defuzzification processing to obtain the prior probability of the node state; inferring the new energy vehicle roll-on / roll-off transportation risk level based on the Bayesian network and each of the prior probabilities.
[0012] In a possible implementation, generating the target prediction probabilities of the respective probability levels by synthesizing the prediction probabilities of the respective probability levels of the node states includes: determining the target prediction probabilities of the respective probability levels of the node states through the following formula:
[0013] wherein, represents the target prediction probability that the node state is at the probability level i , represents the average of the upper and lower bounds of the probability level k .
[0014] In a possible implementation, calculating the triangular fuzzy number of the node state according to the target prediction probabilities of the respective probability levels of the node state includes: determining the average value of the target prediction probabilities and the variance of the target prediction probabilities according to the target prediction probabilities of the respective probability levels of the node state; determining the range of the target prediction probability when in a preset confidence interval according to the average value of the target prediction probabilities and the variance of the target prediction probabilities; determining the triangular fuzzy number of the node state according to the range of the target prediction probability of the node state.
[0015] In a possible implementation, inferring the risk level of new energy vehicle ro-ro transportation according to the Bayesian network and the respective prior probabilities includes: calculating the conditional probabilities of the respective child nodes in the Bayesian network according to the influence probability data between the respective indicators and the Leaky Noisy-Max model; inferring the risk level of new energy vehicle ro-ro transportation based on the prior probabilities of the root nodes and the conditional probabilities of the child nodes.
[0016] In a possible implementation, the method further includes: setting the prediction probability of the highest probability level of the new energy vehicle ro-ro transportation risk level node in the Bayesian network to 100%, and performing backward inference according to the Bayesian network to obtain the posterior probabilities of the respective node states of the respective root nodes; determining the key root nodes according to the difference between the posterior probabilities of the respective node states of the respective root nodes and the prior probabilities.
[0017] In a possible implementation, the risk assessment indicators for new energy vehicle ro-ro transportation include: a plurality of assessment indicators determined according to three dimensions of the fire influence factors, fire disposal management, and risk consequences during new energy vehicle ro-ro transportation.
[0018] In a second aspect, the present invention further provides a risk assessment device for new energy vehicle ro-ro transportation, including: A Bayesian network construction module, which is used to construct a Bayesian network according to the relationships among the evaluation indicators of the ro-ro transportation risk of new energy vehicles and the relationships between the evaluation indicators and the risk levels; An expert judgment result acquisition module, which is used to obtain the first judgment results of experts on the occurrence probabilities of the various node states of the root nodes in the Bayesian network; the first judgment results include: the target probability level and the confidence index for the target probability level; A predicted probability determination module, which is used to use the product of the expert weight and the confidence index as the predicted probability of the target probability level; A new energy vehicle ro-ro transportation risk level determination module, which is used to infer the new energy vehicle ro-ro transportation risk level according to the Bayesian network and the predicted probability.
[0019] The beneficial effects of the present invention are: In summary, by constructing a Bayesian network, the present invention can effectively integrate the evaluation indicators of the ro-ro transportation risk of new energy vehicles, and based on this Bayesian network, the influence of each evaluation indicator on the ro-ro transportation risk level of new energy vehicles can be systematically analyzed. In addition, when determining the predicted probability of the node state of the root node in the Bayesian network through the expert investigation method, considering that it is usually difficult for experts to give an accurate predicted probability of the occurrence probability of the node state, the present invention adopts the method of dividing probability levels to improve the reliability of expert judgment. At the same time, by combining the confidence index of the judgment by experts and the expert weight, the predicted probability of the node state is calculated, further improving the reliability of the predicted probability of the node state. Description of the Drawings
[0020] Figure 1 It is a schematic flowchart of an embodiment of the new energy vehicle ro-ro transportation risk assessment method provided by the present invention; Figure 2 It is a structural diagram of a Bayesian network for the ro-ro transportation risk of new energy vehicles provided by the present invention; Figure 3 It is a schematic flowchart of another embodiment of the new energy vehicle ro-ro transportation risk assessment method provided by the present invention; Figure 4 It is a schematic structural diagram of an embodiment of the new energy vehicle ro-ro transportation risk assessment device provided by the present invention. Detailed Embodiments
[0021] Next, the technical solutions in the embodiments of the present invention will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present invention. Obviously, the described embodiments are only a part of the embodiments of the present invention, rather than all the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative efforts fall within the protection scope of the present invention.
[0022] In the description of the embodiments of the present invention, unless otherwise specified, the meaning of "a plurality of" is two or more. In the embodiments of the present invention, the "first", "second", etc. are used to distinguish similar objects, rather than to describe a specific order or sequence, nor to indicate or imply their relative importance or implicitly specify the quantity of the indicated technical features. It should be understood that such data can be interchanged under appropriate circumstances so that the embodiments of the present application can be implemented in an order other than those illustrated or described herein, and the objects distinguished by "first", "second", etc. are usually of the same category, and the number of objects is not limited. For example, the first object can be one or more.
[0023] Referring to
[0024] In this context, referring to Figure 1 , a flowchart of an embodiment of the new energy vehicle ro-ro transportation risk assessment method provided by the present invention is shown. The method includes: S101, constructing a Bayesian network according to the relationships among the evaluation indicators of the new energy vehicle ro-ro transportation risk and the relationship between the evaluation indicators and the risk levels.
[0025] The safety risk assessment indicators of new energy vehicle ro-ro transportation can be obtained according to the relevant literature on new energy vehicle fire accidents and ro-ro ship navigation safety, new energy vehicle accident reports, and semi-open interview results. Each evaluation indicator is used to determine the risk level of new energy vehicle ro-ro transportation.
[0026] Specifically, the new energy vehicle ro-ro transportation risk assessment indicators may include a plurality of evaluation indicators determined according to three dimensions: the fire impact factors, fire disposal management, and risk consequences during new energy vehicle ro-ro transportation.
[0027] The influencing factors of fire can be analyzed from three aspects: the disaster-forming environment, the disaster-causing factors, and the disaster-bearing bodies. The following 25 evaluation indicators can be obtained. Among them, the disaster-forming environment includes wind force, sea state level, sudden severe convective weather, vehicle cabin temperature, vehicle cabin humidity, vehicle cabin ventilation effect, vehicle cabin drainage effect, and vehicle lashing and securing strength; the disaster-causing factors include proficiency in boarding / disembarking and driving, standardization of getting off the vehicle operation, crew's ship handling ability, vehicle age, vehicle battery state of charge, integrity of the power battery system, power battery temperature, motor system state, vehicle age modification, state and performance of ship electrical equipment, and on-vehicle flammable, explosive and prohibited items; the secondary indicators included in the disaster-bearing bodies are the fire resistance grade of bulkheads and decks, fire resistance performance of ship equipment, sectional loading, parking distance of vehicles in the cabin, and number of people on board.
[0028] The fire disposal management can be analyzed from three aspects: engineering technology, education and training, and safety management. The following 11 evaluation indicators can be obtained. Among them, engineering technology includes completeness of safety inspection equipment, completeness of monitoring and early warning equipment, completeness of isolation facilities and equipment, and pertinence of fire-fighting facilities and equipment; education and training includes comprehensiveness of disaster prevention and mitigation education and training content, frequency of disaster prevention and mitigation education and training; safety management includes safety production responsibility system, completeness of safety operation procedures, completeness of emergency plans, and frequency of emergency drills.
[0029] The measurement indicators of risk consequences can include 5 evaluation indicators: possibility of battery thermal runaway, degree of economic loss, degree of casualties, degree of environmental pollution, and public trust.
[0030] The embodiments of the present invention only exemplarily list some evaluation indicators. Those skilled in the art can also extend the evaluation indicators according to the above evaluation indicator determination method, and no further examples will be given here.
[0031] After obtaining the evaluation indicators, the causal relationships between the evaluation indicators and the causal relationships between the evaluation indicators and the risk level of energy vehicle roll-on / roll-off transportation can be expressed in the form of a directed acyclic graph to obtain a Bayesian network.
[0032] Refer to Figure 2 , which shows a Bayesian network structure diagram of the risk of new energy vehicle roll-on / roll-off transportation provided by the present invention. The Bayesian network includes multiple nodes, and the nodes include index nodes and also risk level nodes (risk level nodes of new energy vehicle roll-on / roll-off transportation). The nodes are connected by directed line segments.
[0033] Each node includes multiple node states. As shown in Table 1, the node states of some nodes are shown. For example, the wind force node includes 5 states: level 0-2, level 3-4, level 5-6, level 7-8, and level 9-11.
[0034] Table 1, Schematic diagram of Bayesian network node states
[0035] S102, obtain the first judgment result of the occurrence probability of each node state of the root nodes in the Bayesian network; the first judgment result includes: the target probability level and the confidence index for the target probability level.
[0036] It is usually difficult for experts to give an accurate probability value for the occurrence probability of node states. Therefore, in this embodiment, a method of dividing probability levels is adopted to improve the reliability of estimation. Specifically, multiple probability levels can be divided according to the size of the occurrence probability. Experts only need to judge which probability level the occurrence probability of the node state of the root node in the Bayesian network belongs to, that is, give a target probability level. At the same time, a confidence index , indicating the degree of recognition of the expert for the judgment result, and its value range can be between 0.5 and 1. When approaches 1, the expert has a higher degree of recognition for the judgment result given by himself.
[0037] Referring to Table 2, a probability level division table provided by the present invention is shown. According to the size of the case occurrence probability, a total of 5 probability levels are divided, and the probability interval of each probability level can be represented by a fuzzy number wherein, is the average value of the interval .
[0038] Table 2, Probability Level Division Table
[0039] S103, use the product of the expert weight and the confidence index as the predicted probability of the target probability level.
[0040] After each expert judges the occurrence probability of each node state of each root node, the product of the expert weight and the confidence index can be used as the predicted probability of the target probability level. Thus, for each node state (up to the Figure 3 embodiment shown at present, the node state refers to the node state of the root node), there are multiple target probability levels obtained by expert judgment and the predicted probabilities corresponding to the target probability levels.
[0041] Referring to Table 3, a table showing the expert judgment results of a root node X33 provided by the present invention is presented. There are 6 experts, and the root node X33 has 5 node states. Taking the judgment result of expert 1 on node state 1 as an example, the confidence index 0.8 corresponding to probability level 3 (the target probability level) is multiplied by the weight 1 of expert 1, and the predicted probability of probability level 3 is obtained as 0.8. By analogy, the target probability levels obtained by each expert's judgment on the occurrence probability of each node state, and the predicted probabilities corresponding to the target probability levels can be obtained. For example, node state 1 corresponds to 6 target probability levels, and the predicted probabilities corresponding to these 6 target probability levels.
[0042] Table 3, Table of Expert Judgment Results for Root Node X33
[0043] Referring to Table 4, a table showing the division of expert weights provided by the present invention is presented. According to the working years, positions, educational backgrounds, etc. of the experts, the expert weights are divided into 5 levels, and the weight of each level is represented by ξ. The values of ξ are 1, 0.9, 0.8, 0.7, and 0.6 respectively. The larger ξ is, the higher the credibility of the expert.
[0044] Table 4, Table of Expert Weight Division
[0045] S104, Infer the roll-on / roll-off transportation risk level of new energy vehicles based on the Bayesian network, target probability level, and predicted probability.
[0046] Each node state corresponds to the target probability levels obtained by each expert's judgment, and the predicted probabilities corresponding to the target probability levels. Therefore, based on statistical methods, statistical analysis of the data of each node state can obtain the prior probability of each node state.
[0047] Then, collect the influence probability data between each node index through an expert questionnaire, and the conditional probabilities of each sub-node in the Bayesian network can be statistically obtained according to this influence probability data.
[0048] Finally, based on the Bayesian network, the prior probabilities of the root nodes and the conditional probabilities of the sub-nodes in the Bayesian network, forward inference can be performed to obtain the roll-on / roll-off transportation risk level of new energy vehicles.
[0049] Referring to Table 5, a table showing the division of the roll-on / roll-off transportation risk level of new energy vehicles provided by the present invention is presented. The roll-on / roll-off transportation risk level of new energy vehicles is divided into 5 risk levels: low risk, relatively low risk, medium risk, relatively high risk, and high risk.
[0050] Table 5, Table of Division of Roll-on / Roll-off Transportation Risk Level of New Energy Vehicles
[0051] The new energy vehicle roll-on / roll-off transportation risk assessment method provided in this embodiment can be applied to a new energy vehicle roll-on / roll-off transportation risk assessment system, and the new energy vehicle roll-on / roll-off transportation risk assessment system can be a software system running on a terminal device. The terminal device can be a tablet computer, a vehicle-mounted device, an augmented reality (AR) / virtual reality (VR) device, a notebook computer, an ultra-mobile personal computer (UMPC), a netbook, a personal digital assistant (PDA), a mobile phone, or other terminal devices. This embodiment does not impose any restrictions on the specific type of the terminal device.
[0052] In summary, in this embodiment, by constructing a Bayesian network, the new energy vehicle roll-on / roll-off transportation risk assessment indicators can be effectively integrated, and based on this Bayesian network, the influence of each assessment indicator on the new energy vehicle roll-on / roll-off transportation risk level can be systematically analyzed. In addition, when determining the prediction probability of the node state of the root node in the Bayesian network through the expert survey method, considering that it is usually difficult for experts to give an accurate prediction probability of the occurrence probability of the node state, so this embodiment adopts the method of dividing probability levels to improve the reliability of expert judgment. At the same time, combining the confidence index of the expert's judgment and the expert weight, the prediction probability of the node state is calculated, further improving the reliability of the prediction probability of the node state.
[0053] In some embodiments of the present invention, S103 includes: determining the prediction probability that the occurrence probability of the node state is in other probability levels according to the probability distribution characteristics and the prediction probability of the target probability level; wherein, the probability distribution characteristics are a normal distribution characteristic with the prediction probability of the target probability level as the center and the prediction probabilities of the probability levels on both sides decreasing; inferring the new energy vehicle roll-on / roll-off transportation risk level according to the Bayesian network and the prediction probabilities of each probability level.
[0054] After obtaining the prediction probabilities of each probability level of the node state, statistical analysis can be performed on the prediction probabilities of each probability level based on statistical methods to obtain the prior probability of the node state, and finally, according to the Bayesian network, the prior probability of the root node in the Bayesian network, and the conditional probability of the child node, forward inference is performed to obtain the new energy vehicle roll-on / roll-off transportation risk level. Among them, the process of obtaining the conditional probability can refer to the above.
[0055] In this embodiment, the predicted probabilities of the occurrence probabilities of node states in other probability levels are supplemented by probability distribution characteristics, and then the ro-ro transportation risk level of new energy vehicles is inferred based on the Bayesian network and the predicted probabilities of each probability level, which can reduce the influence of expert subjective judgment on the determination of the risk level.
[0056] In some embodiments of the present invention, there are multiple experts; the predicted probability is determined by the following formula:
[0057]
[0058]
[0059] Wherein, represents the predicted probability that the occurrence probability of the node state is in the probability level i , represents the predicted probability that the expert n judges that the node state is in the target probability level k , represents the lower bound of the first probability level, represents the upper bound of the first probability level and the lower bound of the second probability level, represents the upper bound of the second probability level and the lower bound of the third probability level, represents the upper bound of the third probability level and the lower bound of the fourth probability level, represents the upper bound of the fourth probability level and the lower bound of the fifth probability level, represents the upper bound of the fifth probability level.
[0060] Taking the data in Table 3 as an example, according to expert 1, the probability level of the occurrence probability of node state 1 is probability level 3, and at the same time, the predicted probability of probability level 3 is calculated to be 0.8. Then the above formula can be applied to calculate the predicted probabilities when the probability levels of the occurrence probability of node state 1 are probability levels 1, 2, 4, and 5 respectively. According to this method, 6 predicted probabilities can be calculated for each of the 5 probability levels of node state 1.
[0061] In some embodiments of the present invention, referring to Figure 3 , the steps of inferring the ro-ro transportation risk level of new energy vehicles according to the Bayesian network and the predicted probabilities of each probability level include: S301, synthesize the predicted probabilities of each probability level of the node state to generate the target predicted probability of each probability level.
[0062] Continuing with the above example, each of the five probability levels of node state 1 corresponds to six predicted probabilities respectively. Statistical analysis can be performed on the six predicted probabilities corresponding to each probability level based on statistical methods to obtain the target predicted probability corresponding to that probability level.
[0063] Specifically, the target predicted probabilities of each probability level of the node state can be determined by the following formula:
[0064] where, represents the target predicted probability that the node state is at the probability level i , and represents the average of the upper and lower bounds of the probability level k .
[0065] S302. After calculating the triangular fuzzy number of the node state based on the target predicted probabilities of each probability level of the node state, defuzzification processing is performed to obtain the prior probability of the node state.
[0066] Continuing with the above example, the node state 1 corresponds to the target predicted probabilities of five probability levels respectively. Similarly, statistical analysis can be performed on the target predicted probabilities corresponding to the five probability levels of the node state 1 based on statistical methods to obtain the triangular fuzzy number of the node state 1. Then, defuzzification processing is performed on the triangular fuzzy number through methods such as the mean area method to obtain the prior probability of the node state 1.
[0067] S303. Based on the Bayesian network and each prior probability, the risk level of new energy vehicle ro-ro transportation is inferred.
[0068] In some embodiments of the present invention, S302 includes: determining the average value of the target predicted probability and the variance of the target predicted probability according to the target predicted probabilities of each probability level of the node state; determining the range of the target predicted probability distribution when in a preset confidence interval according to the average value of the target predicted probability and the variance of the target predicted probability; determining the triangular fuzzy number of the node state according to the range of the target predicted probability distribution of the node state.
[0069] Specifically, a statistical sample is constructed from the predicted probabilities . Since the population variance of the sample is unknown, the sample variance is selected to replace , then the constructed statistical variable follows t distribution. represents 's mean.
[0070] Based on the given , it can be obtained In the confidence interval Probability . It can take a value of 5%.
[0071] root node in The triangular fuzzy number of the state is The defuzzification is performed by the mean area method to obtain the accurate probability value. , that is, the node status The prior probability of . 、 、 are numerically equal.
[0072] In some embodiments of the present invention, S303 includes: calculating the conditional probability of each child node in the Bayesian network based on the influence probability data between each indicator and the LeakyNoisy-Max model; and inferring the risk level of roll-on / roll-off transportation of new energy vehicles based on the prior probability of the root node and the conditional probability of the child node.
[0073] In the Bayesian network model, when multiple parent nodes Affect the same child node together Y When , the model can be converted into a causal Noisy-Max model by introducing the auxiliary node Z. Y It must be an ordered variable. A state value, the value is , the larger the value, the higher the risk level. Parent node It does not need to be a sequential variable, the intermediate node Z Status and X The nodes are consistent, and their status values are determined by the maximum impact of each cause node.
[0074] For the link , let the relevant parameters be ,in, To be the parent node In state When the child node Y Promote to status y The probability of . Define the cumulative conditional probability , then the child node Y The state probability Each parameter in the conditional probability table It can be calculated by the following formula:
[0075] The Leaky Noisy-Max adds a "leakage" probability term on the basis of the standard Noisy-Max. Add variables represents risk factors not incorporated into the model, through represents the impact on the child node Y by the reason not explicitly incorporated, where is the state node when there is no risk, and its cumulative parameter is . The conditional probability calculation formula is updated to:
[0076] Then, based on the prior probability of the node state of the root node and the conditional probabilities of each child node. Or rather, the conditional probabilities between each node, the forward inference can be carried out using GeNIE software to obtain the risk level of the new energy vehicle roll-on / roll-off transportation. The inference formula is:
[0077] In the formula, represents the joint probability distribution of all nodes, represents the set of parent nodes of node , represents the set of parent nodes of node , is the conditional probability of node .
[0078] In some embodiments of the present invention, the new energy vehicle roll-on / roll-off transportation risk assessment method further includes: setting the predicted probability of the highest probability level of the new energy vehicle roll-on / roll-off transportation risk level node in the Bayesian network to 100%, and performing reverse inference according to the Bayesian network to obtain the posterior probability of each node state of each root node; determining the key root nodes according to the difference between the posterior probability and the prior probability of each node state of each root node.
[0079] Specifically, the reverse inference can be carried out according to the following formula:
[0080] In the formula, represents that the new energy vehicle roll-on / roll-off transportation is at the t th risk level; represents that the i rd root node is in the j th state; represents the joint probability of the risk factors of the i th root node; represents the conditional probability that when the state of the i rd root node is , the new energy vehicle roll-on / roll-off transportation is at the risk level .
[0081] Then, the influence degree of each evaluation index on the ro-ro transportation risk is measured by the rate of variation (ROV), and the ROV calculation formula is as follows:
[0082] In the formula, 、 and respectively represent the ROV value, posterior probability, and prior probability of node .
[0083] In summary, the present invention constructs a comprehensive and systematic risk assessment index system for new energy vehicle ro-ro transportation, covering multiple dimensions such as ro-ro transportation environment, personnel operation, new energy vehicle status, ship structure, and emergency management. Secondly, the fuzzy multi-state Bayesian network is used as the evaluation method, which can effectively integrate objective historical data and expert evaluation results, improving the comprehensiveness and reliability of the data; the conditional probability of the intermediate node is corrected by the Leaky Noisy-Max model, simplifying the acquisition form of the conditional probability. Finally, the risk level of new energy vehicle ro-ro transportation is solved by combining the inference algorithm of the Bayesian network model; this method can reduce the subjectivity of risk factor identification, improve the accuracy of risk prediction, and reveal the causal relationship between risk indicators, providing a theoretical basis for shipping enterprises to formulate prevention and control countermeasures for new energy vehicle ro-ro transportation fire accidents.[[ID=:19]]
[0084] Referring to Figure 4 , a schematic structural diagram of an embodiment of the risk assessment device for new energy vehicle ro-ro transportation provided by the present invention is shown. The device includes: A Bayesian network construction module 401, configured to construct a Bayesian network according to the relationship between the evaluation indicators of the new energy vehicle ro-ro transportation risk and the relationship between the evaluation indicators and the risk level; An expert judgment result acquisition module 402, configured to acquire a first judgment result of the probability of occurrence of each node state of the root node in the Bayesian network by an expert; the first judgment result includes: a target probability level and a confidence index for the target probability level; A predicted probability determination module 403, configured to use the product of the expert weight and the confidence index as the predicted probability of the target probability level; A new energy vehicle ro-ro transportation risk level determination module 404, configured to infer the new energy vehicle ro-ro transportation risk level according to the Bayesian network and the predicted probability.
[0085] It should be noted that: the implementation principle or implementation process of the above modules can refer to the embodiments of the new energy vehicle ro-ro transportation risk assessment method described above, and will not be elaborated here one by one.
[0086] Those skilled in the art can understand that all or part of the processes for implementing the methods of the above embodiments can be completed by instructing relevant hardware through a computer program, and the program can be stored in a computer-readable storage medium. Among them, the computer-readable storage medium is a disk, an optical disc, a read-only memory or a random access memory, etc.
[0087] As described above, only the preferred specific embodiments of the present invention are provided, but the protection scope of the present invention is not limited thereto. Any changes or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed by the present invention should be covered by the protection scope of the present invention.
Claims
1. A risk assessment method for the roll-on / roll-off transportation of new energy vehicles, characterized in that Including: Construct a Bayesian network according to the relationships among the evaluation indicators for the ro-ro transportation risk of new energy vehicles and the relationship between the evaluation indicators and the risk levels; Obtain the first judgment results of experts on the occurrence probabilities of the various node states of the root nodes in the Bayesian network; The first judgment results include: the target probability level and the confidence index for the target probability level; Take the product of the expert weight and the confidence index as the predicted probability of the target probability level; Infer the ro-ro transportation risk level of new energy vehicles based on the Bayesian network, the target probability level, and the predicted probability; 2. The risk assessment method for ro-ro transportation of new energy vehicles according to claim 1, wherein The inferring the ro-ro transportation risk level of new energy vehicles based on the Bayesian network, the target probability level, and the predicted probability includes: Determine the predicted probabilities of the occurrence probabilities of the node states being in other probability levels according to the probability distribution characteristics and the predicted probability of the target probability level; wherein, the probability distribution characteristics are a normal distribution characteristic with the predicted probability of the target probability level as the center and the predicted probabilities of the probability levels on both sides decreasing; Infer the ro-ro transportation risk level of new energy vehicles based on the Bayesian network and the predicted probabilities of the various probability levels; 3. The new energy vehicle roll-on / roll-off transportation risk assessment method according to claim 2, wherein There are multiple experts, and there are 5 probability levels; the predicted probability is determined by the following formula: Among them, represents the occurrence probability of the node state being at the probability level i as the predicted probability, represents the expert n judging that the node state is at the target probability level k as the predicted probability, represents the lower bound of the first probability level, represents the upper bound of the first probability level and the lower bound of the second probability level, represents the upper bound of the second probability level and the lower bound of the third probability level, represents the upper bound of the third probability level and the lower bound of the fourth probability level, represents the upper bound of the fourth probability level and the lower bound of the fifth probability level, represents the upper bound of the fifth probability level.
4. The risk assessment method for the roll-on / roll-off transportation of new energy vehicles according to claim 3, wherein The inferring the ro-ro transportation risk level of new energy vehicles based on the Bayesian network and the predicted probabilities of the various probability levels includes: Integrate the predicted probabilities of the various probability levels of the node state to generate the target predicted probabilities of the various probability levels; After calculating the triangular fuzzy number of the node state according to the target predicted probabilities of the various probability levels of the node state, perform defuzzification processing to obtain the prior probability of the node state; Infer the ro-ro transportation risk level of new energy vehicles based on the Bayesian network and the various prior probabilities; 5. The risk assessment method for ro-ro transportation of new energy vehicles according to claim 4, characterized in that, The integrating the predicted probabilities of the various probability levels of the node state to generate the target predicted probabilities of the various probability levels includes: Determine the target predicted probabilities of the various probability levels of the node state by the following formula: Among them, represents the node state at the probability level i of the target prediction probability, represents the average of the upper and lower bounds of the probability level k of.
6. The risk assessment method for ro-ro transportation of new energy vehicles according to claim 4, characterized in that The calculating the triangular fuzzy number of the node state according to the target predicted probabilities of the various probability levels of the node state includes: Determine the average value of the target predicted probability and the variance of the target predicted probability according to the target predicted probabilities of the various probability levels of the node state; Determine the range of the target predicted probability distribution when in the preset confidence interval according to the average value of the target predicted probability and the variance of the target predicted probability; Determine the triangular fuzzy number of the node state according to the range of the target predicted probability distribution of the node state; 7. The risk assessment method for ro-ro transportation of new energy vehicles according to claim 4, wherein, The inferring the ro-ro transportation risk level of new energy vehicles based on the Bayesian network and the various prior probabilities includes: Calculate the conditional probabilities of the various sub-nodes in the Bayesian network according to the influence probability data among the various indicators and the Leaky Noisy-Max model; Based on the prior probability of the root node and the conditional probabilities of the sub-nodes, infer the ro-ro transportation risk level of the new energy vehicle; 8. The risk assessment method for ro-ro transportation of new energy vehicles according to claim 4, wherein The method further includes: Set the predicted probability of the highest probability level of the new energy vehicle ro-ro transportation risk level node in the Bayesian network to 100%, and perform reverse inference according to the Bayesian network to obtain the posterior probabilities of the various node states of each root node; Determine the key root nodes according to the difference between the posterior probabilities and the prior probabilities of the various node states of each root node.
9. The risk assessment method for the roll-on / roll-off transportation of new energy vehicles according to claim 1, wherein The new energy vehicle ro-ro transportation risk assessment indicators include: a plurality of assessment indicators determined according to three dimensions of fire influencing factors, fire disposal management, and risk consequences during new energy vehicle ro-ro transportation.
10. A risk assessment device for ro-ro transportation of new energy vehicles, characterized in that, Include: A Bayesian network construction module for constructing a Bayesian network according to the relationship between the assessment indicators of the new energy vehicle ro-ro transportation risk and the relationship between the assessment indicators and the risk level; An expert judgment result acquisition module for acquiring a first judgment result of the expert on the occurrence probability of each node state of the root node in the Bayesian network; The first judgment result includes: a target probability level and a confidence index for the target probability level; A predicted probability determination module for using the product of the expert weight and the confidence index as the predicted probability of the target probability level; A new energy vehicle ro-ro transportation risk level determination module for inferring the new energy vehicle ro-ro transportation risk level according to the Bayesian network and the predicted probability.