A short-period multi-round situational awareness reliability evaluation method considering anchoring effect and confirmation bias

Abstract

The application provides a short-period multi-round situational awareness reliability evaluation method considering anchoring effect and confirmation bias, and the steps are as follows: step one: obtaining related knowledge of abnormal state, analyzing possible reasons of the current abnormal state, and determining corresponding solutions of each possible reason; step two: qualitative modeling, considering dynamic property of short-period multi-round SA and round correlation caused by anchoring effect and confirmation bias, and modeling the process of short-period multi-round SA by using dynamic Bayesian network; step three: quantitative modeling, determining node parameters in the dynamic Bayesian network on the basis of constructing the structure of the dynamic Bayesian network; and step four: short-period multi-round SA reliability evaluation, evaluating the reliability of the short-period multi-round SA according to the evidence and the determination of the conditional probability table of each node. The application solves the problem that the round correlation caused by anchoring effect and confirmation bias is not considered in the previous short-period multi-round SA reliability evaluation.

Description

TECHNICAL FIELD

[0001] The present application provides a short-period multi-round situation awareness (SA) reliability evaluation method considering anchoring effect and confirmation bias, which analyzes and describes the SA round correlation caused by anchoring effect and confirmation bias, and their mutual promotion and adverse effects on the establishment of correct SA, and models and evaluates the reliability of dynamic short-period multi-round SA, belonging to the field of SA reliability modeling and evaluation. BACKGROUND

[0002] With the development of automation technology, the role of operators has changed from manipulators to supervisors, planners and solvers in abnormal states, in which SA is crucial to decision-making. Scholars from the Department of Industrial Engineering at Texas Tech University, Endsley, described SA as the perception of environmental elements, the understanding of their meaning, and the prediction of their recent state. The present application adopts this definition and defines SA in abnormal states as the perception of current abnormal related information, the understanding of abnormal causes and the prediction of the effectiveness of selected solutions. Establishing SA in abnormal states requires testing and adjusting solutions in multiple rounds of interaction. The failure of SA is a major threat to system safety, according to Endsley's statistics, 88% of aviation accidents are related to the loss of SA in abnormal states. For example, the Boeing 737-8(MAX) accident and the Air France 447 crash were both related to the SA failure of pilots. Therefore, effective and accurate reliability evaluation of multi-round SA is crucial to improving the safety level of human-machine systems.

[0003] In abnormal states, the establishment of SA is characterized by multiple rounds of interaction at short time intervals. The present application mainly studies this short-period multi-round SA. It is different from long-period multi-round SA, which relies on experience and knowledge accumulation. In the process of establishing short-period multi-round SA, the knowledge structure of operators is unlikely to change over time. The cognition of feedback from past actions is the key to correcting false SA. Feedback can trigger more useful evidence to support the exclusion of false options and encourage operators to consider more possible system states. However, cognitive biases, including anchoring effect and confirmation bias, will hinder the correction process. Specifically, anchoring effect will make operators' judgments of system state biased towards previous opinions, leading to insufficient belief adjustment of false SA in the current round; confirmation bias will direct attention to information that confirms previous opinions, leading to the neglect of key information in the current round. For example, in the Boeing 737-8(MAX) accident, these biases led pilots to fail to correct SA in time. Anchoring effect and confirmation bias can promote each other, making false SA beliefs stronger, and SA in each round is not isolated. Therefore, it is important to include the round correlation caused by anchoring effect and confirmation bias in the reliability evaluation of short-period multi-round SA.

[0004] Quantitative methods for SA reliability assessment have been extensively studied. Some studies focus on quantifying the impact of static and dynamic performance shaping factors on SA reliability by extending human factors analysis methods. Other studies assess SA reliability by modeling the dynamic SA establishment process. For example, Kim and Seong from the Department of Nuclear and Quantum Engineering, Korea Advanced Institute of Science and Technology, proposed a mental model-based approach to analyze SA of nuclear power plant operators. They further assessed the impact of attention, memory, and mental model on SA using this approach. Li et al. from the School of Artificial Intelligence and Automation, Huazhong University of Science and Technology, quantitatively assessed SA failure by using dynamic Bayesian networks to consider the dynamic and time-dependent nature of cognitive states. In addition, Stroeve et al. from the Netherlands Aerospace Centre proposed a mathematical model to describe the SA establishment and updating process. They incorporated this model into a multi-agent modeling approach to assess SA reliability. However, these methods are not suitable for assessing the reliability of short-period multi-round SA due to the lack of consideration of action feedback.

[0005] To address this issue, more and more studies have turned their attention to the feedback effect to assess the reliability of dynamically updated short-period multi-round SA. For example, Zhao and Smidts from the Department of Mechanical and Aerospace Engineering, Ohio State University, modeled the cognitive process using Bayesian inference, with feedback activating more system-related knowledge as the number of rounds increases. They proposed an "imperfect" belief propagation algorithm to describe the uncertainty in knowledge retrieval and estimated the reliability of short-period multi-round SA through Monte Carlo simulation. Other studies focus on the case where feedback-triggered new evidence is used for SA updating. Zhang et al. from the School of Civil Aviation, Nanjing University of Aeronautics and Astronautics, used visual and auditory alerts caused by the previous round of operation as new evidence to assess the reliability of pilots' short-period multi-round SA based on Bayesian networks. The latest version of the "Information, Decision, and Action in the Cockpit" model developed a new semantic network-based reasoning module that can simulate operators' knowledge-based reasoning to update SA after action feedback.

[0006] Although the above methods have successfully assessed the reliability of short-period multi-round SA, they ignore the round dependence caused by anchoring effect and confirmation bias. Empirical studies and mechanism studies provide theoretical guidance to overcome this problem. They qualitatively illustrate the impact of confirmation bias-induced dependence on information perception and the impact of anchoring effect-induced dependence on belief updating. However, there is still a gap in the methods for quantifying and modeling these round dependence relationships in quantitative short-period multi-round SA reliability assessment. SUMMARY

[0007] (1) Objectives:

[0008] The application provides a short-period multi-round situational awareness reliability evaluation method considering anchoring effect and confirmation bias.

[0009] (2) Technical scheme:

[0010] The application provides a short-period multi-round situational awareness reliability evaluation method considering anchoring effect and confirmation bias, which comprises four steps of acquiring related knowledge of abnormal state, qualitative modeling, quantitative modeling and short-period multi-round SA reliability evaluation. The method models the dynamic establishment of short-period multi-round SA by using dynamic Bayesian network. The dynamic Bayesian network uses the network in a single time slice to describe the cognitive activity in a single interactive round. The round correlation caused by anchoring effect and confirmation bias is explicitly modeled as the connection between adjacent time slices. Then, the short-period multi-round SA reliability is evaluated based on the dynamic Bayesian network. The method plays an important role in improving personnel SA reliability and enhancing the safety of human-machine systems.

[0011] The application provides a short-period multi-round situational awareness reliability evaluation method considering anchoring effect and confirmation bias, and the flow thereof is as shown in Figure 1 The specific steps are as follows:

[0012] Step one: acquire related knowledge of abnormal state, analyze possible causes of the current abnormal state, and determine the corresponding solution for each possible cause;

[0013] Step two: qualitative modeling, consider the dynamics of short-period multi-round SA and the round correlation caused by anchoring effect and confirmation bias, and model the establishment process of short-period multi-round SA by using dynamic Bayesian network;

[0014] Step three: quantitative modeling, determine the node parameters in the dynamic Bayesian network based on the construction of the dynamic Bayesian network structure;

[0015] Step four: short-period multi-round SA reliability evaluation, evaluate the reliability of short-period multi-round SA according to the evidence and the determination of the conditional probability table of each node;

[0016] Through the above steps, the dynamic establishment of short-period multi-round SA is modeled by using dynamic Bayesian network, the network in a single time slice is used to describe the cognitive activity in a single interaction round, the round correlation caused by anchoring effect and confirmation bias is explicitly modeled as the connection between adjacent time slices, the determination of conditional probability table is guided by the quantification method of round correlation, and finally the reliability of short-period multi-round SA is evaluated based on dynamic Bayesian network, so that the problem that the round correlation caused by anchoring effect and confirmation bias is not considered in the previous reliability evaluation of short-period multi-round SA is solved.

[0017] In step one, the "acquisition of relevant knowledge of abnormal state" is the basis for the construction of the entire dynamic Bayesian network, and the main purpose is to acquire system fault knowledge and human-machine interaction task knowledge for the research object, and to identify the possible causes of the current abnormal state and the subsystem / component faults leading to these causes and the corresponding solutions for each possible cause based on the system fault knowledge and human-machine interaction task knowledge, including the following contents:

[0018] The construction of dynamic Bayesian network depends on the relevant knowledge of abnormal state. These knowledge can be divided into system fault knowledge and human-machine interaction task knowledge. The system fault knowledge includes the possible causes of the current abnormal state and the subsystem / component faults leading to these causes. The possible fault causes have similar fault symptoms, which may cause the operator to confuse the actual cause with other causes. The operator often infers the actual cause according to the cognition of the fault state of these subsystems / components. The human-machine interaction task knowledge contains the information needed to identify the actual cause, and the corresponding solution for each possible cause. This information is related to the abnormal cause, or can indicate the fault state of the subsystem / component.

[0019] In step two, "qualitative modeling" is considered for the dynamics and round correlation of short-period multi-round SA, and a dynamic Bayesian network is used to establish a short-period multi-round SA model under abnormal state, including the following steps:

[0020] Step 1): Identify key nodes in dynamic Bayesian network

[0021] As shown in Figure 2 The method uses a set of nodes to model the perception of each key information, including "evidence" nodes, "attention" nodes and "perception" nodes. Different SAs correspond to different "action" nodes, and the consequences are represented by "utility" nodes. In addition, "anchor" nodes are identified in the dynamic Bayesian network to reflect the round correlation caused by anchoring effect and confirmation bias. These nodes will be expanded and given specific meanings when applied to actual situations. However, the scope of node identification must be limited to prevent the inclusion of information and knowledge unrelated to the abnormal state;

[0022] Step 2) : Determination of the state of all nodes

[0023] This sub-step is to identify the state of the node according to the possible mode of the node, and usually define its state according to the possible operating state of the system function related node, such as failure or success;

[0024] Step 3) : Determination of the relevance within a single round

[0025] The relevance between nodes is established on a single time slice of the dynamic Bayesian network to describe the cognitive activities in a single interactive round, including the establishment of SA and the evaluation of utility. The establishment of the relevance arc in the SA establishment process starts from the observation of the relevant nodes, then the "system state" node, and finally the corresponding "action" node. In the utility evaluation, the development of the relevance arc is similar to the SA establishment process. The above nodes and arcs describe the process of the operator's understanding and handling of the abnormal state in a single round of interaction;

[0026] Step 4) : Establishment of the relevance between adjacent rounds

[0027] This sub-step aims to establish the connection between different dynamic Bayesian network slices, which reflects the round relevance. The most notable round relevance is caused by anchoring effect and confirmation bias.

[0028] The connection between the "anchor" node and the "SA" node is used to reflect the relevance caused by anchoring effect. The relevance caused by confirmation bias can be described by the connection between the "anchor" node of the previous round and the "attention" node of the current round. In addition, the perception of discrete information (such as alarms) is related to their perception state in the previous round. And, the SA of adjacent rounds is relevant. In Figure 2 , these relevance are represented by arrows pointing to their own "perception" nodes and "SA" nodes. The above nodes and connections are combined with the Bayesian network slices to describe the establishment of SA in a short period of multiple rounds.

[0029] Among them, the "quantitative modeling" described in step three is to determine the parameters in the dynamic Bayesian network on the basis of constructing the structure of the dynamic Bayesian network, including the parameter quantization of the nodes with round relevance influence and the nodes without round relevance influence, including the following steps:

[0030] Step 1) : Determination of the state of each round "evidence" node

[0031] The "Evidence" node represents the information that the operator perceives about the updated SA at each interaction round. The identification of its state depends on the trend of the observable variables. These observable variables include discrete information (e.g. alarms) and continuous information (e.g. system states such as aircraft altitude). The use of continuous information in a Dynamic Bayesian Network requires its discretization. The result of the discretization also reflects the operator's subjective estimate of these continuous variables.

[0032] Step 2) : Determination of the conditional probability tables of the nodes without round dependency

[0033] The objective of this sub-step is to determine the conditional probability tables of the system state dependent nodes without considering the dependency induced by anchoring and confirmation bias. The conditional probability tables represent the operator's knowledge and mental model of the abnormal states and can be determined based on their previous training and experience. If the operator is well trained and has sufficient knowledge, these conditional probability tables can be obtained from the system failure logic or standard operating procedures. Otherwise, a questionnaire survey of the operators and expert judgment are also effective methods.

[0034] Step 3) : Quantification of the dependency induced by anchoring and confirmation bias

[0035] This sub-step aims to determine the conditional probability tables of the nodes related to anchoring and confirmation bias in the Dynamic Bayesian Network. To parameterize the anchoring dependent nodes (i.e. the "SA" node), it is necessary to first determine the initial assessment of the evidence set X. It represents the causal strength of the evidence set supporting the anchor without considering the influence of anchoring. Therefore, it can be obtained by referring to the determination of the conditional probability tables of the nodes without cognitive bias. On this basis, the evidence is classified according to the reference point R. Then, the conditional probability tables of the "SA" node are determined using Equations (1) and (2).

[0036]

[0037] where, and P(SA = 1 | X, R = r) and P(SA = 0 | X, R = r) represent the conditional probability of SA being consistent with the initial anchor in the rth round when the operator believes and does not believe the anchor in the (r-1)th round, respectively; X represents the perceived evidence set; X is the evidence set; A represents the anchor; P(X) represents the initial causal strength of the evidence set X supporting the anchor; R is the reference point, which can be set to 0.5; and a and β (0 < a, β < 1) are constants reflecting the operator's sensitivity to negative and positive evidence, respectively.

[0038] In addition, it should be noted that the process of establishing short-cycle multi-round SA is a sequential decision process. When the correct SA is established in a round, the abnormal state can be successfully resolved. Only when the accurate SA is not reached in all the rounds that have occurred, the short-cycle multi-round SA is wrong.

[0039] In the anchor-based situation assessment process, the relevance caused by confirmation bias affects the information perception. This effect determines the probability of information being perceived and is represented by the "attention" node in the dynamic Bayesian network model. Therefore, quantifying the relevance caused by confirmation bias is to calculate the conditional probability table of the "attention" node. First, the salience of the information i in the rth round is calculated using equation (3) (i.e. ). Then, the value of the information in different anchor states is determined according to equations (4) and (5) (i.e. ). Finally, substituting and into equations (6) and (7) can calculate the conditional perception probability of information i.

[0040]

[0041] In the formula, represents the salience of each information in the rth round; and represent the salience based on the change rate of information i, whether information i is an alarm signal, and the human-machine interface design, respectively; γ c , γ a and γ d are the weights of the three salience parameters.

[0042]

[0043] In the formula, is the value of information i in the rth round; is the support of information i to the anchor in the r-1th round.

[0044]

[0045] In the formula, and represent the value of information i in the rth round under the condition that the operator believes and does not believe Anchor r-1 , respectively.

[0046]

[0047] In the formula, A ri is the activation of information i in the rth round; γ and λ are the weights of salience and value.

[0048]

[0049] In the formula, represents the round correlation caused by confirmation bias; and τ and s respectively represent the activation threshold and the parameter of the noise affecting perception.

[0050] In step four, the "short cycle multi-round SA reliability evaluation" is determined according to the evidence and the conditional probability table of each node, the joint probability distribution of the dynamic Bayesian network is calculated, and finally the reliability of the short cycle multi-round SA is obtained, including the following contents:

[0051] According to the evidence and the determination of the conditional probability table of each node, the dynamic Bayesian network realizes the quantitative modeling of the short cycle multi-round SA, and the joint probability distribution can be calculated as follows:

[0052]

[0053] In the formula, P(Z 1:Rn ) is the joint distribution; is the node i in the interactive round r; is the parent node set of , which may contain the node in the rth round or the (r-1)th round; n is the number of nodes in a single round slice; and Rn is the total number of interactive rounds. Therefore, the reliability of the short cycle multi-round SA can be expressed as:

[0054]

[0055] (3) Efficacy, advantage

[0056] The application provides a short cycle multi-round SA reliability evaluation method based on a dynamic Bayesian network, which can obtain abnormal state related knowledge of a research object, identify possible reasons for a current abnormal state and corresponding subsystem / component faults and corresponding solutions, model the short cycle multi-round SA by using the dynamic Bayesian network by considering round correlation, quantify round correlation caused by anchoring effect and confirmation bias, effectively and accurately evaluate the reliability of the short cycle multi-round SA, and improve the safety level of a man-machine system. BRIEF DESCRIPTION OF DRAWINGS

[0057] Figure 1 It is the overall framework of the method.

[0058] Figure 2 It is a schematic diagram of the short cycle multi-round SA based on the dynamic Bayesian network.

[0059] Figure 3The working principle of a maneuvering characteristic enhancement system and related systems.

[0060] Figure 4 A dynamic Bayesian network structure for Boeing 737-8(MAX) accident.

[0061] Figure 5 Reliability of short-cycle multi-round SA in 20 interaction rounds.

[0062] Figure 6 Reliability of short-cycle multi-round SA under different anchoring levels.

[0063] The sequence numbers, symbols and codes in the figures are explained as follows:

[0064] Figure 2 And Figure 4 In the figure, the white thin solid line ellipse box is the "evidence" node, the white dashed line ellipse box is the "attention" node, the white thick solid line ellipse box is the "perception" node, the black ellipse box is the "system state and SA" node, the gray thin solid line ellipse box is the "action" node, the gray dashed line ellipse box is the "utility" node, and the gray thick solid line ellipse box is the "anchor" node; the thick dashed line represents the correlation caused by anchoring effect, the thin dashed line represents the correlation caused by confirmation bias, the thick solid line represents the connection within a single round, and the thin solid line represents the correlation between other rounds.

[0065] Figure 3 In the figure, the white rectangular box represents the structure and function of the system; the gray rectangular box represents visual warning; and the black rectangular box represents auditory warning.

[0066] Figure 5 In the figure, the square icon line and the triangular icon line respectively represent the reliability of short-cycle multi-round SA considering and not considering round correlation; Figure 6 In the figure, the circular icon line, the square icon line and the triangular icon line respectively represent the reliability of short-cycle multi-round SA under high, medium and low anchoring levels. DETAILED DESCRIPTION

[0067] The present application provides a short-cycle multi-round situational awareness reliability evaluation method considering anchoring effect and confirmation bias; the method is performed in the following four steps in sequence, as shown in Figure 1 The present application takes the Boeing 737-8(MAX) accident as the background, and models the dynamic establishment of short-cycle multi-round SA by using the above-mentioned reliability evaluation method, aiming at the multi-round interaction process between the pilot and the aircraft after the abnormal operation of the maneuvering characteristic enhancement system; the model considers the round correlation caused by anchoring effect and confirmation bias, and the determination of the conditional probability table is guided by the quantification method of round correlation, so as to verify the feasibility and accuracy of the present application.

[0068] The specific implementation scheme is described in detail as follows:

[0069] Step 1: Acquire the relevant knowledge of abnormal state

[0070] According to the accident report, we extracted the system failure knowledge and human-machine interaction task knowledge of the Boeing 737-8(MAX) accident. Figure 3 The working principle of the enhanced maneuvering characteristics system and related systems is shown, which represents the system failure knowledge.

[0071] Human-machine interaction task knowledge contains information that needs to be perceived and solutions to the current abnormality. When the delivered parameters are abnormal, the warning system will trigger various visual warnings ( Figure 3 gray rectangular box) and auditory warnings ( Figure 3 black rectangular box). These warnings and indication parameters can help the pilot determine the actual cause of the current abnormality. For the abnormal operation of the enhanced maneuvering characteristics system, the solution is to cut off the horizontal stabilizer automatic trim, and for the failure of the speed trim system, the solution is to manually operate the angle of attack trim.

[0072] Step 2: Qualitative modeling

[0073] Step 1): Identify key nodes in dynamic Bayesian networks

[0074] Dynamic Bayesian networks identify cognitive activities caused by abnormal operation of the enhanced maneuvering characteristics system and knowledge of abnormal state as nodes. For information perception, each warning and indication parameter (such as speed and altitude) is represented by a set of observation-related nodes in the dynamic Bayesian network. In addition, the operation knowledge and failure knowledge of the enhanced maneuvering characteristics system related systems are identified as "system-related" nodes and "action" nodes. The "SA" node representing the pilot's judgment of the cause of the incident is selected as the target node in each round to reflect the reliability of short-period multi-round SA. In addition, inherent nodes (such as "utility" nodes and "anchor" nodes) are also set to reflect the SA belief adjustment caused by action feedback.

[0075] Step 2): Determine the state of all nodes

[0076] For this case, the cause of the accident is the abnormal operation of the enhanced maneuvering characteristics system, which may be misjudged by the pilot as a failure of the speed trim system. Therefore, they are identified as two states of the "SA" node, representing correct SA and incorrect SA, respectively. Each SA state corresponds to an "action" node. For correct SA and incorrect SA, they are cut off horizontal stabilizer automatic trim and manual operation of angle of attack trim, respectively.

[0077] Table 1 lists other nodes and their states.

[0078] Table 1 describes the nodes in the dynamic Bayesian network

[0079]

[0080]

[0081] Step 3): Determine the correlation within a single round

[0082] In a single round, the correlation between the perception information and the system / subsystem such as the "air data module" can be represented by a single fragment of the dynamic Bayesian network, which reflects the pilot's knowledge of the maneuvering characteristics augmentation system related systems. In order to study the reliability of short-period multi-round SA, the focus is on the case of operator anchoring error SA, and only the utility evaluation of the error action that can produce the next round anchor is considered.

[0083] Step 4): Establish the correlation between adjacent rounds

[0084] The round correlation of short-period multi-round SA caused by anchoring effect and confirmation bias can be represented by the connection between the "anchor" node and the "SA" node (thick dashed line in Figure 4 ) and the connection between the "anchor" node and the "attention" node (thin dashed line in Figure 4 ), respectively. As shown in Figure 4 , combined with the internal and inter-time slice connections, the construction of the qualitative dynamic Bayesian network model in the case study can be completed. In Figure 4The states of the "evidence" nodes (terrain warning, "sensed pressure difference" warning, descent rate warning, speed warning, speed inconsistency, altitude warning, altitude inconsistency) and the "attention" nodes (A1-A7) jointly affect the states of the "perception" nodes (terrain warning Per, "sensed pressure difference" warning Per, descent rate warning Per, speed warning Per, speed inconsistency Per, altitude warning Per, altitude inconsistency Per). Only when the state of the "evidence" node is occurrence and the state of the "attention" node is attention, the state of the "perception" node is occurrence. The "perception" nodes affect the "system state" node, and finally affect the state of the "SA" node. According to the state of the "SA" node, the corresponding "action" node is generated. The states of the "attention" nodes (A8-A10) and the "evidence" nodes (angle of attack, altitude, speed) determine the states of the "perception" nodes (angle of attack Per, altitude Per, speed Per), and finally affect the state of the "utility" node. The states of the "action" node and the "utility" node determine the state of the "anchor" node. If the "SA" node correctly judges the system failure cause, the round is terminated. If the "SA" node judges incorrectly, the next round is entered. The round correlation caused by anchoring effect makes the "anchor" node affect the "SA" node of the next round. The round correlation caused by confirmation bias makes the "anchor" node affect the state of the "attention" node (A1-A7). Other round correlations affect some "perception" nodes and the "SA" node of the next round. If the "SA" node always judges incorrectly, the next round will continue until the accident occurs.

[0085] Step three: quantitative modeling

[0086] Step 1): Determine the state of the "evidence" node in each round

[0087] The state of the "evidence" node in each interaction round is determined by the measurement data. The visual and auditory warnings are shown in Table 2.

[0088] Table 2: Warnings existing in each interaction round

[0089]

[0090] Step 2): Determine the conditional probability table of the node without round correlation

[0091] According to the system failure logic and expert judgment, the conditional probability table of the "system / subsystem correlation" node is determined. The conditional probability table with probability logic is selected to encode the pilot's knowledge. Taking the conditional probability table of the "SA" node in the first round as an example, it is shown in Table 3.

[0092] Table 3: Conditional probability table of "SA" node (r=1)

[0093]

[0094] S: success, F: failure

[0095] Step 3): Quantifying the correlation caused by anchoring effect and confirmation bias

[0096] The reasoning process of pilots building SA is supported by new evidence and guided by anchors. Pilots subjectively adjust the initial assessment of evidence strength due to anchoring effect factors. Considering the influence of the previous SA's anchoring effect and anchors, the conditional probability table of the "SA" node can be adjusted as shown in Table 4.

[0097] Table 4 only gives the conditional probability under the condition that the SA in the r-1 round is wrong, i.e. SA[r-1]=FOS. Under the condition that the SA in the r-1 round is correct, the conditional probability is

[0098] Table 4 Conditional probability table of "SA" node (r>1) with anchoring effect correlation

[0099]

[0100] S: success, F: failure; initial anchor = failure of speed trim system

[0101] In each round of interaction, the pilot in the cockpit of the aircraft will get different evidence. However, due to the existence of confirmation bias, the pilot may not be able to perceive all the evidence. The parameters for calculating the conditional probability of each "attention" node related to confirmation bias are shown in Tables 5 and 6.

[0102] Table 5 Parameters for quantifying the correlation caused by confirmation bias

[0103]

[0104]

[0105] Table 6 Salience and value of confirmation bias related evidence

[0106]

[0107] Step four: Short-cycle multi-round SA reliability assessment

[0108] After determining the parameters of all nodes, the short-cycle multi-round SA reliability in the case is calculated using equations (8) and (9).

[0109] In this case, the short-cycle multi-round SA reliability considering and not considering the correlation of SA rounds is Figure 5It can be found that the probability of establishing correct SA is very low in the initial round. This is mainly due to the lack of training of the system-related knowledge of the maneuvering feature enhancement system, unclear description of the abnormal checklist, and missing of key information. Moreover, with the appearance of the key information of the "feeling pressure difference" alarm in the 7th round, the reliability of SA has a mutation, no matter whether the anchoring effect and confirmation bias are considered. Obviously, without considering the round correlation, the mutation degree and the reliability of SA are much higher.

[0110] In addition, the short-period multi-round SA reliability under other anchoring levels, including high anchoring effect and low anchoring effect, is evaluated and compared. The comparison results are shown in FIGS. 6-8. Figure 6 The conclusions under other anchoring levels are similar to those under the medium anchoring level. For example, the addition of the "feeling pressure difference" alarm causes a mutation, reducing the influence of anchoring effect and confirmation bias. The difference is that, with the decrease of the anchoring level, the mutation degree and the short-period multi-round SA reliability are significantly increased.

Claims

1. A short-period multi-round situational awareness reliability evaluation method considering anchoring effect and confirmation bias, characterized in that: The steps are as follows: Step one: acquire the relevant knowledge of the abnormal state, analyze the possible causes of the current abnormal state, and determine the corresponding solution for each possible cause; Step two: qualitative modeling, considering the dynamic nature of short-cycle multi-round SA and the anchoring effect and confirmation bias causing round correlation, using dynamic Bayesian network to model the process of short-cycle multi-round SA; Step three: quantitative modeling, based on the construction of dynamic Bayesian network structure, determine the node parameters in the dynamic Bayesian network; Step four: short-cycle multi-round SA reliability evaluation, according to the evidence and the determination of the conditional probability table of each node, evaluate the reliability of short-cycle multi-round SA; In step two, the dynamic Bayesian network is used to model the process of short-cycle multi-round SA, including the following steps: Step 2.1): identify the key nodes in the dynamic Bayesian network A set of nodes are used to model the perception of each key information, including "evidence" nodes, "attention" nodes and "perception" nodes; different SAs correspond to different "action" nodes, whose consequences are represented by "utility" nodes; in addition, "anchor" nodes are identified in the dynamic Bayesian network to reflect the anchoring effect and confirmation bias causing round correlation; Step 2.2): determine the state of all nodes The state of the node is identified according to the possible mode of the node, and the state of the node is defined according to the possible operating state of the system function related node, including failure or success; Step 2.3): determine the correlation within a single round The correlation between nodes is established in a single time slice of the dynamic Bayesian network, which describes the cognitive activities in a single interactive round, including the establishment of SA and utility evaluation; the establishment of correlation arc in the process of SA establishment starts from the observation of related nodes, then the "system state" node, and finally the corresponding "action" node; Step 2.4): establish the correlation between adjacent rounds Connect different dynamic Bayesian network slices to reflect the round correlation; the most significant round correlation is caused by anchoring effect and confirmation bias.

2. The short-period multi-round situational awareness reliability evaluation method considering anchoring effect and confirmation bias according to claim 1, characterized in that: In step one, the construction of dynamic Bayesian network depends on the relevant knowledge of abnormal state; It is divided into system failure knowledge and human-computer interaction task knowledge; system failure knowledge includes possible causes of the current abnormal state and subsystem / component failures leading to these causes; human-computer interaction task knowledge contains the information needed to identify the actual cause of perception, as well as the corresponding solution for each possible cause; this information is related to the abnormal cause, or indicates the fault state of the subsystem / component.

3. The short-period multi-round situation awareness reliability evaluation method considering anchoring effect and confirmation bias according to claim 1, characterized in that: The connection between "anchor" node and "SA" node is used to reflect the correlation caused by anchoring effect; the correlation caused by confirmation bias is described by the connection between the "anchor" node of the previous round and the "attention" node of the current round; in addition, the perception of discrete information is related to their perception state in the previous round; and the SA of adjacent rounds is related.

4. The short-period multi-round situation awareness reliability evaluation method considering anchoring effect and confirmation bias according to claim 1, characterized in that: In step three, the parameters of the nodes in the dynamic Bayesian network are determined, including the parameter quantification of nodes with round correlation influence and nodes without round correlation influence, the steps are as follows: Step 3.1): Determine the state of the "evidence" node in each round The "evidence" node represents the information that the operator perceives in each interaction round to update the SA; The identification of its state depends on the trend of the observable variables; These observable variables include discrete information and continuous information; the use of continuous information in dynamic Bayesian networks must be discretized; Step 3.2): Determine the conditional probability table of the nodes without round correlation Determine the conditional probability table of the nodes related to the system state without considering the correlation caused by anchoring effect and confirmation bias; the conditional probability table represents the operator's knowledge and mental model of abnormal state, which is determined according to their previous training and experience; if the operator is well trained and has sufficient knowledge, these conditional probability tables are obtained from the system failure logic or standard operating manual; Step 3.3): Quantify the correlation caused by anchoring effect and confirmation bias Determine the conditional probability table of the nodes related to anchoring effect and confirmation bias in the dynamic Bayesian network; to parameterize the anchoring effect related node, i.e. the "SA" node, the initial evaluation of the evidence set X needs to be specified; it represents the causal strength of the evidence set supporting the anchor point without considering the influence of anchoring effect; the evaluation is obtained by reference to the determination of the conditional probability table of the nodes without cognitive bias.

5. The short-period multi-round situation awareness reliability evaluation method considering anchoring effect and confirmation bias according to claim 4, characterized in that: According to the reference point R, the evidence is classified; then use formula (1) and formula (2) to determine the conditional probability table of the "SA" node; wherein and P (SA = r | X, A = r−1) and P (SA = r | X, A = r−1) represent the conditional probability that the SA in the rth round is consistent with the initial anchor point, given that the operator believes and does not believe the anchor point in the r−1th round, respectively; X represents the set of perceptual evidence; X is the set of evidence; A represents the anchor point; P (A | X) represents the initial causal strength that the evidence set X supports the anchor point; R is the reference point, whose value is set to 0.5; α and β, 0 < α, β < 1, are constants that reflect the sensitivity of the operator to negative evidence and positive evidence, respectively.

6. The short-period multi-round situation awareness reliability evaluation method considering anchoring effect and confirmation bias according to claim 4 or 5, characterized in that: The process of establishing short-cycle multi-round SA is a sequential decision-making process; when the correct SA is established in a round, the abnormal state is successfully solved; only when the accurate SA has not been reached all the time in all the rounds that have occurred, the short-cycle multi-round SA is wrong.

7. The short-period multi-round situation awareness reliability evaluation method considering anchoring effect and confirmation bias according to claim 6, characterized in that: In the anchor-based scenario evaluation process, the relevance caused by confirmation bias affects the information perception; this influence determines the probability of information being perceived and is represented by the "attention" node in the dynamic Bayesian network model; therefore, quantifying the relevance caused by confirmation bias is to calculate the conditional probability table of the "attention" node; first, the salience of information i in the rth round is calculated using formula (3), that is Then, according to formula (4) and formula (5), the value of information under different anchor states is determined, that is Finally, substituting and into formula (6) and formula (7) to calculate the conditional perception probability of information i; wherein, represents the saliency of each information in the rth round; and respectively represent the saliency based on the change rate of information i, whether information i is an alarm signal, and the human-machine interface design; γ c , γ a and γ d are the weights of the three saliency parameters; wherein is the value of the information i for the rth round; is the support of the information i for the (r-1)th round anchor. where and denote the value of information i in round r when the operator believes and does not believe Anchor r-1 respectively. where A ri is the activation of the rth round of information i; γ and λ are the weights of salience and value; where denotes the round dependence due to the confirmation bias; τ and s denote the parameters of the activation threshold and the noise impact on perception, respectively.

8. The short-period multi-pass situational awareness reliability assessment method considering anchoring effect and confirmation bias according to claim 1, wherein: In step four, the reliability of short-cycle multi-round SA is evaluated according to the determination of evidence and the conditional probability table of each node, including the following contents: According to the determination of evidence and the conditional probability table of each node, the dynamic Bayesian network realizes the quantitative modeling of short-cycle multi-round SA, and its joint probability distribution is calculated as follows: where P(Z 1:Rn ) is the joint distribution; is the node i in the interaction round r; is the parent node set of , which may contain nodes in the rth round or the r-1th round; n is the number of nodes in a single round slice; Rn is the total number of interaction rounds; therefore, the reliability of the short-period multi-round SA is represented as:

Images