Civil aircraft cockpit display interface availability evaluation index system construction method

Abstract

The invention discloses a civil aircraft cockpit display interface availability evaluation index system construction method, which comprises the following steps: firstly, analyzing pilot human-computer interface interaction cognition through an EPIC model, extracting four-dimensional availability indexes, and preliminarily screening the availability indexes based on the experience and knowledge of a pilot; a civil aircraft cockpit display interface availability evaluation index system is constructed; secondly, in combination with a hierarchical clustering algorithm and a rough set information entropy algorithm, an HC-CEBARKNC algorithm is proposed to perform clustering reduction on the established index system, and a clustering reduction set of indexes of all dimensions is obtained; thirdly, verifying the reliability of each dimension index clustering reduction set by using the classification accuracy of the SVM so as to construct a reasonable civil aircraft cockpit display interface availability evaluation index system; and finally, calculating the weight of each index in the index system by using a rough set conditional information entropy improvement formula. The problem that an index system for civil aircraft cockpit display interface availability research is not comprehensive enough is solved.

Description

Technical Field

[0001] The present invention relates to the field of civil aircraft driving technology, and in particular to a method for constructing a usability evaluation index system for a civil aircraft cockpit display interface. Background Art

[0002] The cockpit display interface of civil aircraft is crucial in civil aviation flight. It is mainly carried by several large-screen color LCD displays. It includes basic information display forms such as numbers, symbols and icons. The main flight display, navigation display, system display, multi-function control display, warning and prompt display provide pilots with systematic and comprehensive flight information and status information to ensure that pilots can successfully perform tasks and make decisions. The cockpit display interface of civil aircraft is an important human-computer interaction interface between pilots and aircraft. With the development of modern civil aviation technology, the flight information and tasks that pilots need to handle are becoming more and more complex, which means that pilots must make effective cognition and decisions in a short time.

[0003] According to aviation accident statistics, pilots' cognitive errors about display interfaces are the main cause of errors. In the human-computer interaction system of civil aircraft cockpits, the usability of display interfaces directly affects the pilots' cognitive accuracy, and thus the performance and safety of the overall flight system. Therefore, the research on the usability of display interfaces has gradually received attention from the aviation industry. At present, based on the interactive cognitive process between pilots and display interfaces, constructing a multi-dimensional civil aircraft cockpit display interface usability evaluation index system has become the main research direction of civil aircraft cockpit display interface usability research.

[0004] The usability evaluation index system is the key to comprehensive evaluation, and building a reasonable evaluation index system is inseparable from index screening. Index screening can reduce information overlap and ensure that each indicator has a significant impact on the evaluation results, thereby improving the efficiency and accuracy of the evaluation.

[0005] The current evaluation index system is mainly developed from two directions: qualitative and quantitative screening: 1. Qualitative index screening is a method that relies on expert opinions or subjective experience; 2. Quantitative index screening methods are to eliminate indicators with overlapping information by analyzing the correlation of indicators. Rough sets, as the most commonly used quantitative index screening method, can keep the information classification unchanged while using attribute reduction methods to mine the information provided by the data itself, simplify knowledge, and use knowledge for decision-making reasoning. Based on the classical rough set theory, a series of attribute reduction algorithms based on conditional information entropy, mutual information, attribute importance, etc. have been proposed in the prior art; and intelligent optimization algorithms such as spider monkey optimization algorithms and genetic algorithms have been proposed inspired by the intelligent behavior of biological groups. However, these intelligent optimization algorithms require a large number of samples and continuous parameter debugging to obtain the reduction results. In practice, it is difficult to obtain a large amount of pilot data, so the above rough set reduction method is not suitable for removing redundant indicators of small sample data.

[0006] Therefore, it is necessary to improve a method for establishing a civil aircraft cockpit display interface usability evaluation index system and an indicator data set screening method suitable for a small sample size and a large number of dimensions to solve the problems existing in the above-mentioned existing technologies. Summary of the invention

[0007] In view of the above-mentioned problems, the present invention constructs a new civil aircraft cockpit display interface usability evaluation index system and an index data set screening method suitable for small sample size and large number of dimensions.

[0008] The technical solution adopted by the present invention is as follows:

[0009] A method for constructing a usability evaluation index system for a civil aircraft cockpit display interface comprises the following steps:

[0010] Step 1: According to the pilot human-machine interface interaction cognitive process based on the EPIC model, multiple dimensions of the civil aircraft cockpit display interface usability indicators are determined, and usability indicators on each dimension are constructed to form an initial usability indicator set for the civil aircraft cockpit display interface;

[0011] Step 2: The pilots preliminarily screen the usability indicators obtained in step 1, obtain a screened usability indicator set, and establish an initial civil aircraft cockpit display interface usability evaluation indicator system;

[0012] Step 3: Use the HC-CEBARKNC algorithm to perform clustering and simplification on the indicators in the availability evaluation index system established in step 2 to obtain a clustering and simplification set of indicators in all dimensions;

[0013] Step 4: Use the SVM classification accuracy to verify the established index clustering reduction set. If the verification passes, proceed to step 6; if the verification fails, return to step 3;

[0014] Step 5: Based on the index clustering reduction set, re-establish the civil aircraft cockpit display interface usability evaluation index system;

[0015] Step 6: Use the rough set conditional information entropy improvement formula to calculate the attribute importance and normalized weight coefficient of each indicator in the usability evaluation index system established in step 6, and use the normalized weight as the weight of each indicator to obtain the final civil aircraft cockpit display interface usability evaluation index system;

[0016] Step 7: Use the civil aircraft cockpit display interface usability evaluation index system in step 6 to evaluate the usability of the civil aircraft cockpit display interface.

[0017] Furthermore, the primary civil aircraft cockpit display interface usability evaluation index system established in step 2 includes i first-level indicators Yi, corresponding to the four dimensions of flight mission, pilot senses, pilot cognition, and flight interaction, and m second-level indicators c i1 ,c i2 ,...,c im Among them, i = 1, 2, 3, 4; m = 1, 2, ..., N, and c im ∈Y i They correspond to the availability indicators retained after the initial screening.

[0018] Furthermore, the specific steps of step 3 include:

[0019] Step 31: Construct the civil aircraft cockpit display interface availability information system S in four dimensions respectively:

[0020] S={U,A,V,f}

[0021] The domain U is a non-empty finite set of objects consisting of pilots; A is a non-empty finite set of attributes that reflect the characteristics of the objects; the information function f:U×A→V is a mapping of attribute values. For any x∈U, a∈A, there is f(x,a)∈V a ; V = ∪ V a is the value range of all attribute values; and A includes independent condition attribute set C and decision attribute set D;

[0022] Step 32: The m secondary indicators c in the primary indicator Yi corresponding to the i-th dimension i1 ,c i2 ,...,c im As the conditional attribute set Ci, n pilots x n The satisfaction of each indicator in Ci is scored independently, and the score is {0, 1, 2}, which means very dissatisfied, average, and very satisfied respectively;

[0023] Step 33: n pilots score the dimensional decision according to the satisfaction score of each indicator in Ci. The score value is {0,1}, which means satisfied or dissatisfied with the decision, respectively. The n score values ​​are used as the decision attribute set Di, Di = {d i1 ,d i2 ,...,d in};

[0024] Step 34: According to the four-dimensional information system S, establish the dimension decision table M of each dimension respectively Yi ;

[0025] Step 35: Update M using agglomerative hierarchical clustering Yi , get the clustered dimension decision table M' Yi ;

[0026] Step 36: Eliminate the dimension decision table M' through the rough set conditional entropy algorithm Yi The redundant indicators that affect the decision attribute set Di are repeated, and a simplified set R of indicators of all dimensions is constructed.

[0027] Furthermore, the specific steps of step 35 include:

[0028] Step 351: Dimension decision table M of each dimension is measured using Euclidean distance Yi The similarity between the availability indicators of the conditional attribute set Ci is used to establish the similarity matrix Ti of each dimension;

[0029] Step 352: cluster Ti based on the Ward hierarchical clustering method to obtain a clustering dendrogram of the availability index of the i-th dimension;

[0030] Step 353: According to the clustering tree diagram, Yi The condition attribute set Ci is divided into k condition attribute subsets A ik , and k ≥ 2;

[0031] Step 354: According to A ik Update dimension decision table M Yi , get the updated dimension decision table M' Yi .

[0032] Furthermore, the specific steps of step 36 include:

[0033] Step 361: Define A ik The equivalence class with Di on U is:

[0034] U / A ik = {Q1,Q2,...,Q r}

[0035] U / Di={P1,P2,...,P e}

[0036] Among them, Q x (x=1,2,…,r) and P s (s=1,2,…,e) are all subsets of U; r and e are A ik The number of equivalence classes with Di on U;

[0037] Step 362: Calculate A ik The information entropy H(A ik ):

[0038]

[0039] Among them, p(Q x )=|Q x | / |U|, means A ik The probability distribution of |Q x | is the set Q x The cardinality of

[0040] Step 363: Calculate A ik The conditional entropy H(Di|A ik ):

[0041]

[0042] Among them, p(Q x |P s )=|Q x ∩P s | / |U|, means A ik The conditional probability distribution of Di;

[0043] Step 364: Calculate the decision attribute set Di relative to A in sequence ik Each conditional attribute c ia The conditional entropy H(Di|{c ia}), sort in descending order according to the calculated conditional entropy value;

[0044] Step 365: Let A ik The initial reduction set is A ik * =A ik , press H(Di|{c ia}) Calculate the decision attribute Di relative index reduction set A in descending order ik * Remove c ia The conditional entropy H'(Di|A ik ):

[0045] H(Di|Aik )=H(Di|A ik * -{c ia});

[0046] Step 366: Determine H'(Di|A ik ) and H(Di|A ik ) are equal, if they are equal, then the conditional attribute cia can be simplified, then Aik * =Aik-{cia}; otherwise, the conditional attribute cia cannot be reduced, Aik * No change; Step 367: Calculate A ik After all the conditional attributes are in, delete the reducible conditional attributes, keep the unreduced conditional attributes, and output the index reduction set A ik * ;

[0047] Step 368: Reduce set A according to the obtained index ik * , construct the reduced set of indicators Yi in each dimension * , Yi * ={A i1 * ∪A i2 * ∪...∪A ik *}, according to the index reduction set Yi of each dimension * Establish a simplified set R of indicators of all dimensions, R = {Y1 * ∪Y2 * ∪Y3 * ∪Y4 *}.

[0048] Furthermore, the specific steps of step 4 include:

[0049] Step 41: The condition attribute set Ci before clustering simplification of the i-th dimension and the condition attribute set Yi after clustering simplification * As the training set of SVM, and the decision attribute set Di as the prediction set of SVM;

[0050] Step 42: Set the conditional attribute sets Yi and Yi * Divide the samples into h parts for cross validation, train the SVM, and obtain the trained SVM classifier;

[0051] Step 43: Use Ci and Yi respectively through the trained SVM classifier * Classify and predict the decision attribute set Di, and calculate Yi and Yi respectively * The classification accuracy η i , ηi * ;

[0052] Step 44: Compare the classification accuracy η of Ci in each dimension i Ci * The classification accuracy η i * , if η is satisfied i <η i * , then the index reduction set after cluster reduction passes the verification; otherwise, it fails the verification and returns to step 3.

[0053] Furthermore, the specific steps of step 6 include:

[0054] Step 61: Reduce the index set A for the i-th dimension ik * , the decision attribute set Di is calculated relative to A by the following formula ik * The conditional information entropy I(Di|A ik * ):

[0055]

[0056] Where t is A ik * The number of equivalence classes on U; U / Aik*={L1,L2,...,Lt} is A ik * The equivalence class on U, L z is a subset of U;

[0057] Step 62: Calculate A by the following formula ik * The conditional information entropy weight ω(A ik * ):

[0058]

[0059] Step 63: For Calculate c by the following formula ia The improved attribute importance NewSig(c ia ):

[0060] NewSig(c ia )=I(Di|A ik * -{c ia})-I(Di|A ik * )+I(Di|{c ia});

[0061] Step 64: Based on NewSig(c ia ) and ω(A ik * ), calculate c ia The final attribute importance Sig(c ia )':

[0062] Sig(c ia )'=NewSig(c ia )ω(A ik * );

[0063] Step 65: Sig(c ia )' is normalized to obtain Sig(c ia )' is the normalized weight

[0064]

[0065] Step 66: Normalize the weights to As the weight value ω(c ia ),Right now The beneficial effects of the present invention are:

[0066] First, starting from the pilot's EPIC cognitive model of the display interface, the present invention preliminarily establishes a multi-dimensional civil aircraft cockpit display interface usability evaluation index system, avoiding one-sided and subjective selection of indicators, and ensuring the comprehensiveness and integrity of the preliminarily established index system;

[0067] Second, based on the hierarchical clustering algorithm and the rough set conditional entropy algorithm, the present invention proposes the HC-CEBARKNC index clustering reduction algorithm, which can screen the index data set with few samples and many dimensions for the usability evaluation index system of the cockpit display interface of civil aircraft;

[0068] Third, the present invention adopts the classification accuracy of the four dimensions after SVM prediction index clustering simplification, and uses the condition attribute set before clustering simplification and the index reduction set after clustering simplification to classify and predict the decision attribute set, respectively, to verify whether the HC-CEBARKNC algorithm deletes redundant indicators while maintaining the original knowledge classification ability of the availability indicators, so as to ensure the reliability and rationality of the establishment of the final civil aircraft cockpit display interface availability evaluation index system. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 This is a flow chart of the indicator evaluation system construction and indicator clustering simplification proposed by the present invention.

[0070] Figure 2 Schematic diagram of the pilot's human-machine interface interaction cognitive process based on the EPIC model.

[0071] Figure 3 Hierarchical clustering dendrogram of availability indicators for flight mission dimensions.

[0072] Figure 4 Hierarchical clustering tree diagram for the availability indicators of pilot sensory dimensions.

[0073] Figure 5 This is a hierarchical clustering tree diagram of the availability indicators of pilot cognitive dimensions.

[0074] Figure 6 Hierarchical clustering dendrogram for flight interaction dimension availability indicators.

[0075] Figure 7 The following is a comparison chart of the SVM accuracy distribution in four dimensions before and after clustering reduction using the HC-CEBARKNC algorithm.

[0076] Figure 8 A civil aircraft cockpit display interface usability evaluation index system constructed for the present invention. DETAILED DESCRIPTION

[0077] In order to enable those skilled in the art to better understand the technical solution of the present invention, the technical solution of the present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0078] The core idea of ​​the present invention is as follows: firstly, the pilots' human-machine interface interaction cognition is analyzed through the EPIC model, and the usability indicators of four dimensions are extracted. The usability indicators are preliminarily screened based on the pilots' experience and knowledge, and a new civil aircraft cockpit display interface usability evaluation index system is constructed; then, based on the hierarchical clustering algorithm and the rough set information entropy algorithm, an HC-CEBARKNC indicator clustering reduction algorithm suitable for data sets with a small number of samples and a large number of dimensions is proposed, and the established indicator system is clustered and reduced to obtain a four-dimensional indicator clustering reduction set; then, the reliability of the indicator clustering reduction set of each dimension is verified by using the classification accuracy of the SVM, and a reasonable civil aircraft cockpit display interface usability evaluation index system is constructed; finally, the weight coefficient of each indicator in the indicator system is calculated based on the rough set conditional information entropy improved formula.

[0079] Example

[0080] The present invention discloses a method for constructing a cockpit display interface usability index system, comprising the following steps:

[0081] Step 1: Preliminary determination of the usability indicators of civil aircraft cockpit display interfaces

[0082] S1: Determine the four dimensions of the usability index of the civil aircraft cockpit display interface, and obtain the usability index of each dimension in the four dimensions;

[0083] The specific steps include:

[0084] S11: Based on Figure 2 The pilot human-machine interface interaction cognitive process based on the EPIC model is shown, and the four dimensions of the usability index of the civil aircraft cockpit display interface are determined: flight mission dimension, pilot sensory dimension, pilot cognitive dimension, and flight interaction dimension;

[0085] The usability index of the flight mission dimension is related to the pilot's mission situation and mission efficiency as well as the task compatibility of the display interface;

[0086] The usability index of the pilot sensory dimension focuses on the visual visibility and auditory accessibility of the pilot to obtain information on the display interface;

[0087] The usability index of the pilot cognitive dimension is related to the pilot's memory and learning;

[0088] The usability indicators of the flight interaction dimension are related to the real-time and predictive nature of the display interface.

[0089] S12: A preliminary set of usability indicators for cockpit display interfaces of civil aircraft is constructed from four dimensions, and the preliminary set of indicators is supplemented and revised by combining the Nielsen model of general product usability evaluation and the software quality model (ISO 9126). The preliminary set of usability indicators for cockpit display interfaces of civil aircraft is obtained as shown in Table 1, with a total of 43 usability indicators.

[0090] Table 1 Preliminary availability indicators

[0091]

[0092] Step 2: Preliminary screening of the initial set of availability indicators

[0093] S2: The pilots conduct a preliminary screening of the initial availability index set established in S1;

[0094] It specifically includes the following steps:

[0095] S21: The usability indicators shown in Table 1 were preliminarily screened by pilots with flight time greater than 30,000 h, 35 usability indicators were retained, and 8 usability indicators were deleted; the deleted indicators included: task accuracy and task completion time in the flight mission dimension; balance, volume adjustability, and audio optimization in the pilot sensory dimension; learnability in the pilot cognitive dimension; and interaction standardization and instant interaction response in the flight interaction dimension.

[0096] S22: The usability evaluation index set is defined as Y = {Y1, Y2, Y3, Y4}, where Y1 represents the flight mission dimension, Y2 represents the pilot sensory dimension, Y3 represents the pilot cognitive dimension, and Y4 represents the flight interaction dimension. The description of the retained usability indicators and indicator symbols after the preliminary screening of the four dimensions of Y1, Y2, Y3, and Y4 is shown in Table 2. The usability evaluation index set Y for the display interface of the civil aircraft cockpit can be described as:

[0097]

[0098] Table 2 Symbols for reserved availability indicators

[0099]

[0100] Step 3: Establish an initial usability evaluation index system based on the selected usability indicators

[0101] S3: Establish an initial usability evaluation index system for the usability of civil aircraft cockpit display interfaces;

[0102] It specifically includes the following steps:

[0103] S31: The first level indicator is defined as Y i (i=1,2,3,4), corresponding to flight mission, pilot sense, pilot cognition, and flight interaction, respectively;

[0104] S32: The secondary indicator is defined as c i1 ,c i2 ,...,c im (i=1,2,3,4;m=1,2,…,N), and c im ∈Y i , c im It represents the availability indicators retained after the initial screening of each dimension in Table 2.

[0105] Step 4: Simplify the availability indicator system to obtain the index clustering reduction set

[0106] S4: Based on the HC-CEBARKNC algorithm, clustering and reducing calculations are performed on the availability evaluation index system initially established in step S3 to obtain a clustering and reducing set of indicators in four dimensions;

[0107] This embodiment takes the flight mission dimension as an example to describe the index clustering and simplification process of the flight mission dimension:

[0108] S41: Construct the civil aircraft cockpit display interface availability information system S of four dimensions respectively, S = {U, A, V, f}, U represents a non-empty finite object set consisting of pilots, A represents a non-empty finite attribute set reflecting the characteristics of the object, including a mutually independent conditional attribute set C and a decision attribute set D, satisfying C∪D=A and C∩D=φ; V=∪V a Represents the range of all attribute values; function f:U×A→V is the mapping of attribute values. For any x∈U, a∈A, there is f(x,a)∈V a ;

[0109] S42: The m availability indicators c included in Yi i1 ,c i2 ,...,c im As the conditional attribute set Ci, invite n pilots x n The availability indicators contained in the condition attribute set Ci are scored independently, and the score values ​​are {0, 1, 2}, where the score 0 indicates that the indicator is very unsatisfactory, the score 1 indicates that the indicator is average, and the score 2 indicates that the indicator is very satisfactory; the data is standardized and discretized through the scoring values;

[0110] S43: The pilot's satisfaction with the usability index of the civil aircraft cockpit display interface in the i-th dimension is taken as the decision attribute set Di, whose value is {0,1}: the decision value 0 indicates dissatisfaction, and the decision value 1 indicates satisfaction;

[0111] S44: Construct dimensional decision table M based on the four-dimensional information system S Yi (i=1,2,3,4), define the conditional attribute set Ci of the i-th dimension={c i1 ,c i2 ,...,c im}, the decision attribute set is Di = {di1, di2, ..., din}, and the domain is n pilots (x n ), that is, U = {x1, x2, ..., xn};

[0112] This embodiment takes the flight mission (i=1) as an example, and constructs the dimension decision table M Y1 , as shown in Table 3.

[0113] Table 3 Dimensional decision table of flight missions

[0114]

[0115]

[0116] S45: Use the agglomerative hierarchical clustering method to analyze the i-th dimension decision table M Yi Update and obtain the clustered dimension decision table M'Yi , the specific steps are:

[0117] 1) Use Euclidean distance to measure the dimension decision table M Yi The similarity between the two availability indicators contained in the conditional attribute set Ci in , and form a similarity matrix Ti;

[0118] 2) Using Ward hierarchical clustering method to cluster the similarity matrix Ti, we can get the clustering tree diagram of the availability index of the i-th dimension. The clustering tree diagram of the flight mission dimension (i=1) is as follows: Figure 3 As shown;

[0119] 3) According to the clustering tree diagram, the decision table M Yi The conditional attribute set Ci is divided into k conditional attribute subsets, k ≥ 2, that is, Ci = {Ai1∪Ai2∪...∪Aik}, and c ia ∈A ik ;

[0120] Taking the flight mission as an example, when k = 3, Yi is divided into Y1 = {A 11 ,A 12 ,A 12}(c 1a ∈A 1k ,), where: A 11 ={c 16 ,c 17 ,c 19}, A 12 ={c 11 ,c 12 ,c 14}, A 13 ={c 13 ,c 15 ,c 110};

[0121] 4) Update the dimension decision table M according to the hierarchical clustering results of the previous step Yi , get the updated dimension decision table M' Yi ;

[0122] The updated dimension decision table of the flight mission is shown in Table 4.

[0123] Table 4 Dimensional decision table of updated flight missions

[0124]

[0125]

[0126] S46: Use the rough set conditional entropy CEBARKNC algorithm to screen the i-th dimension condition attribute set Ci again, and remove the dimension decision table M' in turnYi The redundant indicators that affect the decision attribute set Di are repeated;

[0127] Taking the flight mission as an example, the specific steps include:

[0128] 1) A 11 The equivalence class with D1 on U is U / A 11 ={{x1},{x2},{x3,x4,x6},{x5},{x7}}, the equivalence class derived from D1 is U / D1={{x2,x3,x6},{x1,x4,x5,x7}};

[0129] 2) Calculate A 11 The information entropy H(A 11 )=2.1281;

[0130] 3) Calculate A 11 The conditional entropy H(D1|A 11 )=0.5714;

[0131] 4) Calculate H(D1|{c ia}), each conditional attribute c ia According to H(D1|{c ia The values ​​of}) are arranged in descending order, and the calculation results are shown in Table 5;

[0132] 5) Let A 11 The initial reduction set A 11 * =A 11 , press H(D1|{c ia}) Calculate the decision attribute D1 relative index reduction set A in descending order 11 * Remove c ia The conditional entropy H'(D1|A 11 ), the calculation results are shown in Table 5;

[0133] 6) If H'(D1|A 11 )==H(D1|A 11 ), it means that the conditional attribute c can be simplified ia , at this time A 11 * =A 11 -{cia}; otherwise, the conditional attribute cia cannot be reduced, A 11 * No change; calculate the conditional attribute subset A ik All c ia After that, delete the availability indicator c 16 、c 17 , retain the availability index c 19, the final output index reduction set A 11 * ={c 19};

[0134] Table 5 Conditional entropy calculation results

[0135]

[0136]

[0137] 7) Reduce set A according to the index 11 * , and obtain the reduced set of flight mission dimension indicators Y1 * :

[0138] Y1*={A 11 *∪A 12 *∪A 13 *}={{c 19}∪{c 12}∪{c 13 ,c 15 ,c 110}};

[0139] S47: Repeat step S45 to obtain the hierarchical clustering dendrograms of the usability indicators of the pilot sensory dimension, the pilot cognitive dimension, and the flight interaction dimension, as shown in FIG. Figure 4 , Figure 5 and Figure 6 Repeat step S46 to calculate the index reduction set Yi of the four dimensions of the civil aircraft cockpit display interface usability evaluation index system * , establish a simplified set R of indicators of all dimensions:

[0140]

[0141] Step 5: Verify whether the cluster reduction is reasonable. If not, rebuild the decision table.

[0142] S5: Use the SVM classification accuracy to verify the rationality of the clustering reduction of the four-dimensional indicators. If the SVM classification accuracy verification of a certain dimension is unreasonable, it is necessary to return to step S3 to reconstruct the decision table of each dimension;

[0143] The specific steps include:

[0144] S51: In the i dimension, the condition attribute set Ci before clustering simplification and the condition attribute set Yi after clustering simplification are * As the training set of SVM, the decision attribute set Di is used as the prediction set of SVM;

[0145] S52: To verify the effectiveness of clustering reduction using the HC-CEBARKNC algorithm, the Gaussian radial basis kernel function of the libsvm toolkit is used to train the SVM classifier in MATLAB software, and the conditional attribute set Yi and the index reduction set Yi are transformed into * Divide into h samples for cross validation, the Gaussian radial basis kernel function K(σ i ,σ x )for:

[0146]

[0147] Among them, δ is the scale parameter;

[0148] S53: Use SVM to predict classification accuracy, using the conditional attribute set Yi before clustering simplification and the index reduction set Yi after clustering simplification respectively * Classify and predict the decision attribute set Di;

[0149] The classification accuracy is calculated by the classification accuracy calculation formula, and the classification accuracy η of the conditional attribute set Yi is output. i 、Indicator reduction set Yi * The classification accuracy η i * ;

[0150] The calculation formula of the classification accuracy is:

[0151]

[0152] Among them, h is the total number of samples, h1 is the number of samples whose training results of the training set samples are the same as the actual category, and h2 is the number of samples whose training results of the prediction set samples are the same as the actual category;

[0153] The classification accuracy of the four dimensions is calculated by the above formula:

[0154] η1=51.23,η1 * =76.52(η1<η1 * )

[0155] η2=53.46,η2 * =81.66(η2<η2 * )

[0156] η3=62.18,η3 * =78.29(η3<η3 * )

[0157] η4=58.15,η4 * =80.01(η4<η4 * )

[0158] The SVM accuracy distribution comparison diagram of the four dimensions in this embodiment before and after the HC-CEBARKNC algorithm cluster reduction is as follows: Figure 7 As shown;

[0159] S54: Determine η of each dimension i Whether η is satisfied i <η i * , if satisfied, the clustering reduction is reasonable;

[0160] From the above classification accuracy calculation results, we can see that the η of the four dimensions i Are greater than or equal to η i * ,It is proved that the usability indicators retained after cluster simplification are more reliable and can be used to construct a usability evaluation indicator system for the cockpit display interface of civil aircraft.

[0161] Step 6: Construct a new usability evaluation index system based on index clustering and reduction set transformation

[0162] S6: Based on the four dimensions and the clustering reduction set of indicators in each dimension, establish a usability evaluation index system for the cockpit display interface of civil aircraft;

[0163] It specifically includes the following steps:

[0164] S61: The four dimensions Yi (i = 1, 2, 3, 4) are still used as the first-level indicators, including: flight mission, pilot senses, pilot cognition, and flight interaction;

[0165] S62: Yi after clustering simplification * The indicators included in the report are used as secondary indicators, including: task orientation, satisfaction, task accuracy, information matching, task complexity, interface layout, interface information, interface color, voice warning prompt, understanding, minimum memory load, simplicity, consistency, information real-time, information accuracy, interactive feedback, setting change, and collision warning.

[0166] Based on the above-mentioned primary and secondary indicators, a civil aircraft cockpit display interface usability evaluation index system is constructed. The established evaluation index system is as follows: Figure 8 shown.

[0167] Step 7: Calculate the weight of each indicator in the established evaluation index system and normalize it

[0168] S7: Set the weights of the first-level indicators of the indicator system to be the same, and use the rough set conditional information entropy improved formula to calculate the attribute importance and normalized weight coefficient of each indicator in the indicator system;

[0169] Taking the flight mission dimension as an example, the specific steps include:

[0170] S71: Reduced set A for flight mission dimensions 1k * , the decision attribute set Di relative to A is calculated by the conditional entropy calculation formula 1k * The conditional information entropy I(D1|A 1k * );

[0171] The conditional information entropy calculation formula is:

[0172]

[0173] The calculated conditional information entropy is:

[0174] I(D1|A 11 * )=8 / 49;

[0175] I(D1|A 12 * ) = 2 / 49;

[0176] I(D1|A 13 * ) = 4 / 49;

[0177] S72: Since the conditional attribute Y1 is divided into k conditional attribute sets, A is calculated according to the conditional information entropy weight calculation formula. 1k * The conditional information entropy weight ω(A 1k * );

[0178] The calculation formula of the conditional information entropy weight is:

[0179]

[0180] The calculated weights are:

[0181] ω(A 11 * )=0.5714;

[0182] ω(A 12 * )=0.1429;

[0183] ω(A 13 * )=0.2857;

[0184] S73: For Index c ia The improved attribute importance calculation formula based on rough set information entropy is:

[0185] NewSig(c ia )=I(Di|A ik * -{c ia})-I(Di|A ik * )+I(Di|{c ia});

[0186] The calculated improved attribute importance NewSig(c ia )for:

[0187] NewSig(c 19 )=0.2449;

[0188] NewSig(c 12 )=0.1225;

[0189] NewSig(c 13 )=0.4081;

[0190] NewSig(c 15 )=0.2041;

[0191] NewSig(c 110 )=0.2041;

[0192] S74: For indicators By changing the indicator c ia The improved attribute importance of NewSig(c ia ) and the conditional information entropy weight ω(A 1k * ) is weighted to calculate the final attribute importance Sig(c ia )';

[0193] The final attribute importance calculation formula is:

[0194] Sig(c ia )'=NewSig(c ia )ω(A ik * );

[0195] The final attribute importance calculated is:

[0196] Sig(c 19 )'=ω(A 11 * )×NewSig(c 19 )=0.1399;

[0197] Sig(c 12)'=ω(A 12 * )×NewSig(c 12 )=0.0175;

[0198] Sig(c 13 )'=ω(A 13 * )×NewSig(c 13 )=0.1166;

[0199] Sig(c 15 )'=ω(A 13 * )×NewSig(c 15 )=0.0583;

[0200] Sig(c 110 )'=ω(A 13 * )×NewSig(c 110 )=0.0583;

[0201] S75: According to S74, the final attribute importance Sig(c ia )', since the weights of the four dimensions of the civil aircraft cockpit display interface usability index evaluation system are the same, the final attribute importance Sig(c ia )' perform normalization to obtain any index c ia ∈R’s final attribute importance Sig(c ia )''s normalized weight use The weight values ​​of all indicators in the index reduction set R can be determined, where The calculation formula is:

[0202]

[0203] S76: Set any index c ia The normalized weight of ∈R is As the indicator weight ω(c ia ),Right now The weight values ​​ω(c ia )The calculation results are shown in Table 6.

[0204] Table 6 Calculation results of weight values ​​of indicators

[0205]

[0206] Step 8: Evaluate the usability of the civil aircraft cockpit display interface based on the usability evaluation index system established in step 7.

[0207] The above shows and describes the basic principles, main features and advantages of the present invention. It should be understood by those skilled in the art that the present invention is not limited to the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention to be protected. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A method for constructing a usability evaluation index system for a civil aircraft cockpit display interface, characterized in that: The following steps are involved: Step 1: According to the pilot human-machine interface interaction cognitive process based on the EPIC model, multiple dimensions of the civil aircraft cockpit display interface usability indicators are determined, and usability indicators on each dimension are constructed to form an initial usability indicator set for the civil aircraft cockpit display interface; Step 2: The pilots preliminarily screen the usability indicators obtained in step 1, obtain a screened usability indicator set, and establish an initial civil aircraft cockpit display interface usability evaluation indicator system; Step 3: Use the HC-CEBARKNC algorithm to perform clustering and simplification on the indicators in the availability evaluation index system established in step 2 to obtain a clustering and simplification set of indicators in all dimensions; Step 4: Use the SVM classification accuracy to verify the established index clustering reduction set. If the verification passes, proceed to step 6; If the verification fails, return to step 3; Step 5: Based on the index clustering reduction set, re-establish the civil aircraft cockpit display interface usability evaluation index system; Step 6: Use the rough set conditional information entropy improvement formula to calculate the attribute importance and normalized weight coefficient of each indicator in the usability evaluation index system established in step 6, and use the normalized weight as the weight of each indicator to obtain the final civil aircraft cockpit display interface usability evaluation index system; Step 7: Use the civil aircraft cockpit display interface usability evaluation index system in step 6 to evaluate the usability of the civil aircraft cockpit display interface.

2. A method for constructing a civil aircraft cockpit display interface usability evaluation index system according to claim 1, characterized in that: The primary civil aircraft cockpit display interface usability evaluation index system established in step 2 includes i first-level indicators Yi, corresponding to the four dimensions of flight mission, pilot senses, pilot cognition, and flight interaction, and m second-level indicators c i1 ,c i2 ,...,c im Among them, i = 1, 2, 3, 4; m = 1, 2, ..., N, and c im ∈Y i They correspond to the availability indicators retained after the initial screening.

3. A method for constructing a usability evaluation index system for a civil aircraft cockpit display interface according to claim 2, characterized in that: The specific steps of step 3 include: Step 31: Construct the civil aircraft cockpit display interface availability information system S in four dimensions respectively: S={U,A,V,f} The domain U is a non-empty finite set of objects consisting of pilots; A is a non-empty finite set of attributes that reflect the characteristics of the objects; the information function f:U×A→V is a mapping of attribute values. For any x∈U, a∈A, there is f(x,a)∈V a ; V = ∪ V a is the value range of all attribute values; and A includes independent condition attribute set C and decision attribute set D; Step 32: The m secondary indicators c in the primary indicator Yi corresponding to the i-th dimension i1 ,c i2 ,...,c im As the conditional attribute set Ci, n pilots x n The satisfaction of each indicator in Ci is scored independently, and the score is {0, 1, 2}, which means very dissatisfied, average, and very satisfied respectively; Step 33: n pilots score the dimensional decision according to the satisfaction score of each indicator in Ci. The score value is {0,1}, which means satisfied or dissatisfied with the decision, respectively. The n score values ​​are used as the decision attribute set Di, Di = {d i1 ,d i2 ,...,d in }; Step 34: According to the four-dimensional information system S, establish the dimension decision table M of each dimension respectively Yi ; Step 35: Update M using agglomerative hierarchical clustering Yi , get the clustered dimension decision table M' Yi ; Step 36: Eliminate the dimension decision table M' through the rough set conditional entropy algorithm Yi The redundant indicators that affect the decision attribute set Di are repeated, and a simplified set R of indicators of all dimensions is constructed.

4. A method for constructing a civil aircraft cockpit display interface usability evaluation index system according to claim 3, characterized in that: The specific steps of step 35 include: Step 351: Dimension decision table M of each dimension is measured using Euclidean distance Yi The similarity between the availability indicators of the conditional attribute set Ci is used to establish the similarity matrix Ti of each dimension; Step 352: cluster Ti based on the Ward hierarchical clustering method to obtain a clustering dendrogram of the availability index of the i-th dimension; Step 353: According to the clustering tree diagram, M Yi The condition attribute set Ci is divided into k condition attribute subsets A ik , and k ≥ 2; Step 354: According to A ik Update dimension decision table M Yi , get the updated dimension decision table M' Yi .

5. A method for constructing a civil aircraft cockpit display interface usability evaluation index system according to claim 4, characterized in that: The specific steps of step 36 include: Step 361: Define A ik The equivalence class with Di on U is: U / A ik ={Q1,Q2,...,Q r } U / Di={P1,P2,...,P e } Among them, Q x (x=1,2,…,r) and P s (s=1,2,…,e) are all subsets of U; r and e are A ik The number of equivalence classes with Di on U; Step 362: Calculate A ik The information entropy H(A ik ): Among them, p(Q x )=|Q x | / |U|, means A ik The probability distribution of |Q x | is the set Q x The cardinality of Step 363: Calculate A ik The conditional entropy H(Di|A ik ): Among them, p(Q x |P s )=Q x ∩P s | / |U|, means A ik The conditional probability distribution of Di; Step 364: Calculate the decision attribute set Di relative to A in sequence ik Each conditional attribute c ia The conditional entropy H(Di|{c ia }), sort in descending order according to the calculated conditional entropy value; Step 365: Let A ik The initial reduction set is A ik * =A ik , press H(Di|{c ia }) Calculate the decision attribute Di relative index reduction set A in descending order ik * Remove c ia The conditional entropy H'(Di|A ik ): H(Di|A ik )=H(Di|A ik * -{c ia }); Step 366: Determine H'(Di|A ik ) and H(Di|A ik ) are equal, if they are equal, then the conditional attribute cia can be simplified, then Aik * =Aik-{cia}; otherwise, the conditional attribute cia cannot be reduced, Aik * No change; Step 367: Calculate A ik After all the conditional attributes are in, delete the reducible conditional attributes, keep the unreduced conditional attributes, and output the index reduction set A ik * ; Step 368: Reduce set A according to the obtained index ik * , construct the reduced set of indicators Yi in each dimension * , Yi * ={A i1 * ∪A i2 * ∪...∪A ik * }, according to the index reduction set Yi of each dimension * Establish a simplified set R of indicators of all dimensions, R = {Y1 * ∪Y2 * ∪Y3 * ∪Y4 * }.

6. A method for constructing a civil aircraft cockpit display interface usability evaluation index system according to claim 5, characterized in that: The specific steps of step 4 include: Step 41: The condition attribute set Ci before clustering simplification of the i-th dimension and the condition attribute set Yi after clustering simplification * As the training set of SVM, and the decision attribute set Di as the prediction set of SVM; Step 42: Set the conditional attribute sets Yi and Yi * Divide the samples into h parts for cross validation, train the SVM, and obtain the trained SVM classifier; Step 43: Use Ci and Yi respectively through the trained SVM classifier * Classify and predict the decision attribute set Di, and calculate Yi and Yi respectively * The classification accuracy η i , η i * ; Step 44: Compare the classification accuracy η of Ci in each dimension i Ci * The classification accuracy η i * , if η is satisfied i <η i * , then the index reduction set after cluster reduction passes the verification; otherwise, it fails the verification and returns to step 3.

7. A method for constructing a civil aircraft cockpit display interface usability evaluation index system according to claim 6, characterized in that: The specific steps of step 6 include: Step 61: Reduce the index set A for the i-th dimension ik * , the decision attribute set Di is calculated relative to A by the following formula ik * The conditional information entropy I(Di|A ik * ): Where t is A ik * The number of equivalence classes on U; U / Aik*={L1,L2,...,Lt} is A ik * The equivalence class on U, L z is a subset of U; Step 62: Calculate A by the following formula ik * The conditional information entropy weight ω(A ik * ): Step 63: For Calculate c by the following formula ia The improved attribute importance NewSig(c ia ): NewSig(c ia )=I(Of|A ik * -{c ia })-I(Of|A ik * )+I(Of|{c ia }); Step 64: Based on NewSig(c ia ) and ω(A ik * ), calculate c ia The final attribute importance Sig(c ia )': Mr.(c) ia )'=NewSig(c ia )ω(A ik * ); Step 65: Sig(c ia )' is normalized to obtain Sig(c ia )' is the normalized weight Step 66: Normalize the weights to As the weight value ω(c ia ),Right now

Images