Flight test system configuration design intelligent decision-making method based on business rules

Abstract

The invention discloses a business rule-based test flight test system configuration design intelligent decision-making method, which comprises the following steps of: constructing a test parameter requirement database in which test parameter requirements of different parameter types and specific data of the test parameter requirements in a previous test flight test task are stored; a sensor information base is established, the sensor information base comprises a sensor list for selection, and each sensor corresponds to the sensor attribute index of the parameter type to which the sensor belongs; constructing a sensor selection decision-making tree, training the sensor selection decision-making tree by taking specific data required by the test parameters as training data, and screening selection decision-making rule conditions; the sensor selection scheme meeting the decision rule selection condition is used as an alternative scheme; and determining a comprehensive score of each alternative scheme, and selecting the alternative scheme with the highest comprehensive score as a final scheme. The method has important practical significance and application prospects, and powerful technical support is provided for efficient and high-quality development of the civil aircraft industry.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of flight test system configuration design and artificial intelligence decision-making, and particularly relates to a flight test system configuration design intelligent decision-making method based on business rules. BACKGROUND

[0002] Under the background of the rapid development of the current civil aircraft industry, improving flight test efficiency has become a key task. Flight test system, as an indispensable airborne test facility in aircraft flight test, can collect massive data to verify whether the aircraft state parameters meet the design expectations. However, the flight test system configuration needs to be dynamically changed according to the flight test task requirements of each flight test in the flight test implementation process, and the flight test system is required to be able to respond quickly and be configured flexibly.

[0003] Existing intelligent decision-making methods such as ID3, C4.5, CART, etc. need to first establish a training set. For each test parameter requirement, the expert judges which sensors in the sensor information library can meet the test parameter requirement. Then, the sensor attribute index that can be used as the decision tree judgment rule is trained based on the training set. The decision tree judgment rule only meets the basic constraints and cannot select the optimal solution from multiple sensor alternative schemes that meet the test parameter requirements. Moreover, the lack of systematic induction of expert experience affects the design consistency. SUMMARY

[0004] The purpose of the present application is to provide a flight test system configuration design intelligent decision-making method based on business rules to improve the efficiency of flight test system configuration in the flight test implementation process.

[0005] In order to achieve the above task, the present application adopts the following technical solutions: A flight test system configuration design intelligent decision-making method based on business rules, comprising: Constructing a test parameter requirement database, which stores test parameter requirements of different parameter types and specific data of test parameter requirements in previous flight test tasks; Establishing a sensor information library, which contains a list of selectable sensors, and each sensor has a sensor attribute index corresponding to the parameter type to which the sensor belongs; Building a sensor selection decision tree, training the sensor selection decision tree with the specific data of the test parameter requirements as training data to screen selection decision rule conditions; Selecting sensor selection schemes that meet the selection decision rule conditions as alternative schemes; determining the comprehensive score of each alternative scheme, and selecting the alternative scheme with the highest comprehensive score as the final scheme.

[0006] Further, a sensor selection decision tree is constructed, and specific data of test parameter requirements are taken as training data to train the sensor selection decision tree to screen selection decision rule conditions, including: First, a sensor selection decision tree training set is established, including: finding all test parameter requirements of the parameter type to which the sensor belongs from the test parameter requirement library; and taking these test parameter requirements as labels of the sensor selection decision tree training set. Second, in the test parameter requirement database, specific data of test parameter requirements in previous flight test tasks corresponding to each test parameter requirement are taken as training data of the sensor selection decision tree training set; with the help of expert knowledge, each test parameter requirement in the sensor selection decision tree training set is polled to mark whether each sensor in the sensor information library meets the test parameter requirement, and the judgment result of whether to meet is taken as the marking result of each group of training data in the sensor selection decision tree training set. Finally, based on the ID3 decision tree algorithm, a sensor selection decision tree is constructed by calculating information gain; the sensor selection decision tree contains multiple test parameter requirements screened out by training, and these test parameter requirements are taken as selection decision rule conditions.

[0007] Further, when determining the comprehensive score, first, a judgment rule table of the sensor selection decision tree is constructed, second, expert synthesis weights corresponding to each selection decision rule condition in the judgment rule table are solved, third, single scores corresponding to each selection decision rule condition are solved, and finally, the comprehensive score of each candidate scheme is obtained based on the expert synthesis weights and the single scores.

[0008] Further, a judgment rule table of the sensor selection decision tree is constructed, including: According to the selection decision rule conditions contained in the sensor selection decision tree, corresponding test parameter requirements are found from the test parameter requirement database, and corresponding sensor attribute indexes are found from the sensor information library; according to the correspondence among the selection decision rule conditions, the test parameter requirements, and the sensor attribute indexes, the judgment rule table is constructed.

[0009] Further, the expert synthesis weights corresponding to each selection decision rule condition in the judgment rule table are solved, including: First, for all selection decision rule conditions in the judgment rule table, a best-worst fuzzy comparison matrix is constructed, including: The most important selection decision rule condition and the least important selection decision rule condition are selected from all selection decision rule conditions in the form of expert voting; the first expert's opinion on the importance of the selection decision rule condition is taken as the first expert synthesis weight of the selection decision rule condition ; the first expert's opinion on the importance of the selection decision rule condition is taken as the first expert synthesis weight of the selection decision rule condition . Fuzzy comparison vector , The element in is the first Paired comparisons by experts Compared to other choice decision rules and conditions The importance of the fuzzy weights assigned accordingly; Construct the first An expert on Fuzzy comparison vector , The element in is the first Experts compared the conditions of other choice decision rules with those of individual experts. The importance of each element is assigned a fuzzy weight; all fuzzy weights are represented in triplet form. Then the best-worst fuzzy comparison matrix ; The second step is to define the optimization function and solve the... Each expert analyzed the conditions of the decision-making rule for each choice. Assigned fuzzy weights ; The third step is to utilize the fuzzy weights of different experts. Solving for expert composite weights .

[0010] Furthermore, the optimization function is expressed as follows:

[0011]

[0012] in, Indicates the first One expert Assigning fuzzy weights, Indicates the first One expert Assigning fuzzy weights; Indicates the first One expert The assigned fuzzy weights are represented in triplet form as follows: , These represent the corresponding lower limit, modal value, and upper limit, respectively. Indicates the first Paired comparisons by experts Compared to The importance of the fuzzy weights assigned accordingly; Indicates the first Paired comparisons by experts Compared to The importance of the fuzzy weights assigned accordingly; To use the conversion function GMIR to convert the fuzzy weight to a crisp value; To determine the number of selection decision rule conditions in the rule table.

[0013] Further, in solving the optimization function, the variable is introduced, while , , , , are all expressed in the form of corresponding triples, so that the goal of the optimization problem becomes to minimize the variable , where represents the maximum absolute distance between the ratio of the fuzzy weights assigned by the th expert to and and , and the maximum absolute distance between the ratio of the fuzzy weights assigned by the th expert to and and , to simplify the solving process.

[0014] Further, the fuzzy weights of different experts are used to solve the expert synthetic weight , specifically: The average of the crisp values of the fuzzy weights assigned by all M experts is taken as the expert synthetic weight of the selection decision rule condition.

[0015] Further, the single-item score corresponding to each selection decision rule condition is solved, and finally the comprehensive score of each alternative is obtained based on the expert synthetic weight and the single-item score, including: First, the corresponding sensor attribute index, test parameter requirement, and single-item score relationship are designed according to the characteristics of each selection decision rule condition; the relationship is divided into two cases: one is a piecewise function relationship, and the other is a selection function relationship; Second, the single-item score is solved according to the relationship, and the average of the single-item score weighted by the corresponding expert synthetic weight is taken as the comprehensive score.

[0016] Further, the parameter types include temperature test parameters, pressure test parameters, vibration test parameters, and strain test parameters, and the test parameter requirements of each parameter type remain consistent; The sensor attribute index and the test parameter requirement in the database correspond one-to-one to the content of the test parameter requirement of the corresponding parameter type. ​​

[0017] A terminal device comprises a processor, a memory and a computer program stored in the memory; when the processor executes the computer program, the intelligent decision-making method for flight test system configuration design based on business rules is realized.

[0018] A computer-readable storage medium stores a computer program; when the computer program is executed by a processor, the intelligent decision-making method for flight test system configuration design based on business rules is realized.

[0019] Compared with the prior art, the present application has the following technical features: The method of the present application integrates expert knowledge and AI decision-making technology, breaks through the limitations of low efficiency and easy errors of traditional manual configuration, and realizes automatic and accurate decision-making in the sensor selection link. Through a multi-expert weight fusion mechanism, dispersed industry experience is converted into quantifiable rules, ensuring decision consistency and knowledge inheritance. It supports comprehensive decision-making of multiple test parameters to meet the needs of dynamic flight test tasks, especially suitable for configuration scenarios with high complexity and high safety requirements. Based on the interpretable weight calculation rules (such as fuzzy decision method), a traceable decision path is provided to avoid the risk of traditional AI "black box". The research of the present application in the field of intelligent airborne test system design technology not only has far-reaching academic value, but also has important practical significance and application prospect, providing strong technical support for the efficient and high-quality development of the civil aircraft industry. BRIEF DESCRIPTION OF DRAWINGS

[0020] Figure 1 The decision-making method workflow diagram in an embodiment of the present application. DETAILED DESCRIPTION

[0021] The present application provides an intelligent decision-making method for flight test system configuration design based on business rules, which takes a test parameter demand database and a sensor information library as input, relies on expert knowledge annotation data, constructs a sensor selection decision tree training set, trains a sensor selection decision tree, and obtains a sensor selection decision rule condition. With the help of fuzzy mathematics method, the expert knowledge is converted into fuzzy weight of selection decision rule condition, and a calculation method of expert synthetic weight of sensor selection alternative scheme is further proposed. The calculation relationship between the expert synthetic weight of sensor selection alternative scheme and the comprehensive score of sensor selection alternative scheme is established, and through the sorting of the comprehensive scores of multiple sensor selection alternative schemes, the optimal sensor selection alternative scheme is selected; the specific steps of the present application are as follows: Step 1: Establish a test parameter demand database to provide constraint conditions for sensor selection.

[0022] The test parameter requirement database contains the test parameter requirements needed to complete the flight test mission. According to the parameter type, they can be divided into: temperature test parameters, pressure test parameters, vibration test parameters, strain test parameters, etc. The test parameter requirements for each parameter type are consistent and specifically include: test parameter name, measurement unit, upper and lower limits of measurement range, measurement accuracy, test frequency, wiring method, measurement point installation location, upper and lower limits of measurement point ambient temperature range, sensor volume, measurement point installation size restrictions, and sensor shape restrictions, etc.

[0023] The test parameter requirement database also contains specific data on the test parameter requirements from previous flight test missions, that is, a set of specific values ​​corresponding to each test parameter requirement.

[0024] Step 2: Establish a sensor information database.

[0025] The sensor information database contains a list of available sensors. Each sensor corresponds to a sensor attribute index for its parameter type. These sensor attribute indices correspond one-to-one with the test parameter requirements for the corresponding parameter type in the test parameter requirement database. The sensor attribute indexes can be used to determine whether the sensor meets the test parameter requirements.

[0026] Sensor attribute indicators include the following: (1) Temperature sensor attribute indicators: upper and lower limits of measurement range, measurement accuracy, sensor shape (probe type, hard patch type, flexible patch type, clamp type), sensor size (probe length, patch area length, patch area width, clamp diameter), sensor volume, price; (2) Vibration sensor attribute indicators: measurement range, upper and lower cutoff frequency limits, measurement accuracy, number of vibration axes (single axis, three axes), sensor length, sensor width, sensor height, upper limit of ambient temperature, lower limit of ambient temperature, natural frequency of the measured structure, and price; (3) Strain sensor properties: upper and lower limits of measurement range, measurement accuracy, strain sensor shaft type (single-piece, multi-piece), strain sensor resistance (120Ω, 350Ω), temperature compensation function, temperature compensation material (ferritic steel, aluminum, austenitic steel, silicon dioxide / composite material, titanium / gray cast iron, plastic), upper and lower limits of ambient temperature, price. (4) Pressure sensor attribute indicators: upper and lower limits of measurement range, measurement accuracy, type of measured medium (liquid, gas), upper and lower limits of cutoff frequency, upper limit of ambient pressure, upper and lower limits of ambient temperature, and price.

[0027] Step 3: Build a sensor selection decision tree to filter selection decision rule conditions.

[0028] First, a training set for the sensor selection decision tree is established, including: finding all test parameter requirements for the sensor's parameter type from the test parameter requirement library; and using these test parameter requirements as labels for the sensor selection decision tree training set. Taking a temperature sensor as an example, when constructing the decision tree training set, it is necessary to select all temperature test parameters corresponding to the temperature test parameters, including the test parameter name, measurement unit, upper and lower limits of the measurement range, measurement accuracy, test frequency, wiring method, measurement point installation location, upper and lower limits of the measurement point's ambient temperature range, measurement point installation size limitations, and sensor shape limitations, and use these as labels.

[0029] Secondly, in the test parameter requirement database, the specific data of the test parameter requirements in previous flight test missions corresponding to each test parameter requirement are used as training data for the sensor selection decision tree training set.

[0030] With the help of expert knowledge, each test parameter requirement in the sensor selection decision tree training set is polled, and each sensor in the sensor information database is labeled to determine whether it meets the test parameter requirements. The result of whether it meets the requirements is used as the labeling result of each set of training data in the sensor selection decision tree training set.

[0031] Based on the traditional ID3 decision tree algorithm, a sensor selection decision tree is constructed by calculating information gain. The sensor selection decision tree contains multiple test parameter requirements selected through training, and these test parameter requirements are used as selection decision rule conditions.

[0032] Step 4: Calculate the weights of the alternative solutions for sensor selection.

[0033] In practical engineering applications, all sensor selection schemes that meet the selection rules in the sensor selection decision tree are called candidate schemes. Since there is not a unique candidate scheme, it is necessary to calculate the comprehensive score for each candidate scheme. By combining the scores of all alternative solutions Sort the options to obtain the optimal alternative.

[0034] Step 4.1: Establish the decision rule table for the sensor selection decision tree.

[0035] Based on the selection decision rules contained in the sensor selection decision tree, the corresponding test parameter requirements are found from the test parameter requirement database, and the corresponding sensor attribute indicators are found from the sensor information database. Based on the correspondence between the selection decision rules, test parameter requirements and sensor attribute indicators, a judgment rule table is constructed.

[0036] Step 4.2 Solve for the conditions of each selection decision rule in the decision rule table of the sensor selection decision tree. Corresponding expert synthesis weights .

[0037] In engineering applications, different experts often have different choices and decision-making rules. Given differing perceptions of the importance of each expert's opinion, this solution considers the opinions of every expert. To synthesize the judgment weights of multiple experts, a fuzzy mathematics method is proposed to transform expert knowledge into fuzzy weights for selecting decision rules. Furthermore, this method synthesizes the assigned weights from multiple experts into a single expert composite weight. The specific algorithm is as follows: (1) Let the number of decision rules in the sensor selection decision tree table be 1. The conditions of the choice decision rule are expressed as follows: , For the first Each choice decision rule condition. .

[0038] (2) Note that it needs to be passed through a total of If each expert assigns weights to all the conditions of the choice decision rule, then the... Fuzzy comparison matrix of experts as follows:

[0039] Fuzzy comparison matrix middle, Indicates the first Experts on the conditions for choosing decision rules Compared to another choice decision rule condition The fuzzy weighting results of the importance degree, where .

[0040] (3) In order to obtain the weight assignment results in step (2) Experts used a set of vague linguistic terms To evaluate the conditions for choosing decision rules Relative to the conditions of the selection decision rule The importance of the five fuzzy language terms, their meanings, and the conversion rules for fuzzy weight assignment are shown in Table 1. These five fuzzy language terms are defined, and their weight assignments are represented by triangular membership functions, expressed as triples. ,in , and These are the lower limit, modal value, and upper limit of the membership function of the triangle, respectively.

[0041] Table 1. Meanings of Fuzzy Language Terms and Conversion Rules for Fuzzy Weight Assignment

[0042] (4) Establish the best-worst fuzzy comparison matrix .

[0043] In practical applications, the fuzzy comparison matrix in step (2) is used... When determining the fuzzy weight assignment results for each expert's choice decision rule conditions, it is necessary to consider each choice decision rule condition... Compared with all other selection decision rules, the calculation process is cumbersome, and the selection decision rules that affect the final judgment result are often the most important ones.

[0044] Therefore, based on Rezaei's Best-Worst Method (BWM), we simplified the fuzzy comparison matrix. The best-worst fuzzy comparison matrix is ​​obtained. The specific process is as follows: The decision-making rules and conditions were determined through expert voting. Choose the most important decision rule conditions and the least important choice decision rule conditions ;No. Experts on the most important decision-making rules and conditions Fuzzy comparison vector The definition is as follows:

[0045] in Indicates in total In the conditions of the choice decision rule, the first... The most important selection decision rule conditions are compared in pairs among experts. Compared to other choice decision rules and conditions The importance of each element is assigned a fuzzy weight, which is represented using triples:

[0046] in, , Fuzzy weights The lower bound, modal value, and upper bound of the triangular membership function expression are derived from the first... Each expert assigned values ​​according to the conversion rules for the meaning and weighting of fuzzy language terms in Table 1; and when hour, .

[0047] No. Experts' conditions for the least important selection decision rule Fuzzy comparison vector The definition is as follows:

[0048] in, Experts Pairwise comparison of other choice decision rules conditions Compared to the least important choice decision rule conditions The importance of each element is assigned a fuzzy weight, which is represented using triples:

[0049] in, , Fuzzy weights The lower bound, modal values, and upper bound of the membership function expression of the triangle are determined by experts. Refer to Table 1 for the conversion rules of the meaning and weight assignment of fuzzy language terms; and when hour, .

[0050] Thus, fuzzy comparison vector And fuzzy comparison vector The combination formed the first The best-worst fuzzy comparison matrix of the experts is represented as follows: =( , ).

[0051] Best-Worst Fuzzy Comparison Matrix Provided directly by experts based on experience, as a subsequent step in solving the problem. Each expert analyzed the conditions of the decision-making rule for each choice. Assigned fuzzy weights Input conditions.

[0052] (5) In order to solve the first Each expert analyzed the conditions of the decision-making rule for each choice. Assigned fuzzy weights The optimization function is defined as follows:

[0053]

[0054] The optimization objective is to make the th The most important selection decision-making rule condition is the individual expert. and the Item selection decision rule conditions The ratio of the assigned fuzzy weights Fuzzy weights for pairwise comparisons Minimize the maximum absolute distance; at the same time, make the first The expert for the first Item selection decision rule conditions and the least important choice decision rule conditions The ratio of the assigned fuzzy weights Fuzzy weights for pairwise comparisons Minimize the maximum absolute distance; where .

[0055] In the optimization function above: Indicates the first The most important selection decision-making rule condition is the individual expert. The assigned fuzzy weights are represented using triangular membership functions. , is the unknown to be solved.

[0056] Indicates the first The least important expert is the choice of decision-making rule condition. The assigned fuzzy weights are represented using triangular membership functions. , is the unknown to be solved.

[0057] Indicates the first Experts select decision rule conditions The assigned fuzzy weights are represented using triangular membership functions. , , is the unknown to be solved.

[0058] Indicates the first The most important selection decision rule conditions are compared in pairs among experts. Compared to other choice decision rules and conditions The importance of each fuzzy weight is assigned and expressed using a triangular membership function. The best-worst fuzzy comparison matrix is ​​derived from step (3). , is a known number.

[0059] Indicates the first Each expert compares the conditions of other choice decision rules in pairs. Compared to the least important choice decision rule conditions The importance of each fuzzy weight is assigned and expressed using a triangular membership function. The best-worst fuzzy comparison matrix is ​​derived from step (3). , is a known number.

[0060] because Includes and ,when When the objective calculation formula is equal to 0, this case is not considered in subsequent calculations.

[0061] The above function optimization problem needs to satisfy three constraints simultaneously: Constraint 1: For the normalization verification condition, the first Each expert has a choice of decision-making rule conditions. Assigned fuzzy weights The sum after conversion to simplified values ​​is 1; where, To use the transformation function GMIR to fuzzy weights The result is converted to a concise value; the fuzzy number is then processed using the GMIR function. The formula for converting to a simplified value is:

[0062] Constraint 2: Represents fuzzy weights The corresponding rules governing the relationship between the lower limit, modal value, and upper limit of the membership function of the triangle.

[0063] Constraint 3: This indicates that all selection decision rule conditions are involved in the calculation.

[0064] Introducing variables The optimization function is simplified to:

[0065] The goal of the optimization problem becomes minimizing a variable. , use it to represent the first The most important selection decision-making rule condition is the individual expert. and The ratio of the assigned fuzzy weights Fuzzy weights for pairwise comparisons The maximum absolute distance, and the first An expert became and the least important choice decision rule conditions The ratio of the assigned fuzzy weights Fuzzy weights for pairwise comparisons The maximum absolute distance.

[0066] Experts For the most important decision-making rule conditions Fuzzy weights Other choice decision rule conditions The ratio of the fuzzy weights should be equal to the first... Fuzzy weights for pairwise comparisons provided by experts Try to be as consistent as possible, among which for , , for .

[0067] Indicates the first Each expert group has other choice decision-making rules and conditions. Fuzzy weights The least important choice decision rule conditions Fuzzy weights The ratio should be in line with experts Fuzzy weights for pairwise comparisons Try to be as consistent as possible, among which for , for , for .

[0068] (6) Calculate the expert composite weight .

[0069] For each choice decision rule condition The corresponding fuzzy weights are obtained by solving the optimization function. ; with all M A simplified value for assigning fuzzy weights by an expert. The average value is the condition of the selection decision rule. Expert Synthesis Weight , means as follows:

[0070] Step 4.3, Solve for the conditions of each choice decision rule. Corresponding individual score .

[0071] Based on the information in Table 2, and according to the conditions of each selection decision rule... The characteristic design of the corresponding sensor attribute indicators Test parameter requirements and individual scores The relational expression is divided into two types: piecewise function and selection function. The relational expression can be set according to actual needs to obtain the corresponding individual scores. .

[0072] Step 4.4: Calculate the overall score for each alternative. The final decision is made by comparing all alternatives and selecting the one with the highest overall score.

[0073] For each choice decision rule condition Each has a specific individual score. ,according to The overall score can be calculated. for: .

[0074] Example: One embodiment of the present invention takes a temperature sensor as an example. The selection decision tree for the constructed temperature sensor includes the following selection decision rules: upper and lower limits of measurement range, measurement accuracy, wiring method, sensor size, and sensor shape limitation. The judgment rule table of the constructed temperature sensor selection decision tree is shown in Table 2.

[0075] in, This indicates that the temperature sensor selection decision tree obtained from training... Each choice decision rule has conditions.

[0076] Indicates the first The first selection decision rule condition corresponds to the first condition in the test parameter requirement database. Each test parameter requirement.

[0077] Indicates the first The corresponding decision rule condition in the sensor information database is the first... Sensor attribute indicators.

[0078] Indicates the first Each choice decision rule condition The corresponding expert synthesis weight needs to be calculated in subsequent steps.

[0079] Table 2. Judgment Rules of the Temperature Sensor Selection Decision Tree

[0080] Based on Table 2, assuming that by Experts participated, with a total of Each choice decision rule condition needs to be assigned a weight.

[0081] First, according to step (4), five experts voted to select the winner. The most important selection decision rule condition. The least important choice decision rule condition.

[0082] Secondly, based on the best-worst fuzzy comparison matrix from step (4) In this format, experts 1 through 5, based on their respective experience, directly provide their respective fuzzy judgment matrices. The specific results are as follows:

[0083]

[0084]

[0085]

[0086]

[0087] Then, based on the above data, an optimization equation is constructed: Let the objective function be:

[0088] With the fuzzy judgment matrix of expert 1 For example, the optimization equation is:

[0089] Solving the optimization equation, we get: minimize u The optimal value is: u =0.7481.

[0090] The results of expert 1 assigning weights to the conditions of the 5 selection decision rules are as follows:

[0091] Subsequently, the fuzzy weights of the expert's 1-to-5 choice decision rule conditions were converted into simplified values ​​using the transformation function GMIR. The result is as follows:

[0092]

[0093]

[0094]

[0095]

[0096] Verify the normalization conditions:

[0097] Similarly, the fuzzy weights of the other four experts' assignments to the five selection decision rule conditions were calculated using the above method, and then converted into simplified values ​​to verify the normalization conditions.

[0098] Finally, for each choice decision rule condition The average of the simplified values ​​of the fuzzy weights assigned by all five experts is taken as the expert composite weight for this selection decision rule condition. Finally, the weights of all five decision rule conditions are obtained. ,in:

[0099]

[0100] For the first type of piecewise function relationship, taking Table 2 as an example, the selection decision rule conditions... The upper and lower limits of the measurement range are defined as follows: The score for each item is... The actual upper and lower limits of the sensor's measurement range are taken from the sensor's attribute indicators. The upper and lower limits of the actual test parameter requirement measurement range are taken as the test parameter requirement. In general engineering applications, it is believed that the closer the upper and lower limits of the sensor's measurement range are to the upper and lower limits of the required measurement range of the test parameters, the more fully the sensor's measurement range is being utilized. Simultaneously, the upper and lower limits of the sensor's measurement range should be appropriately larger than the upper and lower limits of the required measurement range of the test parameters to reserve some margin. However, the upper and lower limits of the sensor's measurement range should not be too large compared to the upper and lower limits of the required measurement range of the test parameters; otherwise, a significant portion of the sensor's measurement range will remain unused. This design logic is based on... 、 、 It conforms to the following piecewise function relationship:

[0101] in, , and The coefficient is generally set based on engineering experience. , , .

[0102] Another scenario involves choosing a functional relationship, specifically the decision rule conditions in Table 2. For the wiring method, the score for this item is: Since the wiring method follows a mandatory matching rule and has no range of values, it is a state parameter. Therefore, it can be designed... 、 、 It conforms to the following selection function relationship:

[0103] However, in the actual decision-making process, if This solution cannot be determined through the decision tree, therefore it is the default solution among all alternatives. .

[0104] Similarly, we obtain the individual scores corresponding to all five choice decision rule conditions. .

[0105] The above embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. A smart decision-making method for the configuration design of a flight test system based on business rules, characterized in that, include: Construct a test parameter requirement database, which stores test parameter requirements for different parameter types, as well as specific data on test parameter requirements from previous flight test missions; Establish a sensor information database containing a list of available sensors, with each sensor corresponding to a sensor attribute index for the parameter type to which it belongs; A sensor selection decision tree is constructed, and the specific data of the test parameter requirements are used as training data to train the sensor selection decision tree in order to filter the selection decision rule conditions. Sensor selection schemes that meet the selection decision rule conditions are considered as alternative schemes; Determine the overall score for each alternative and select the alternative with the highest overall score as the final solution.

2. The intelligent decision-making method for flight test system configuration design based on business rules as described in claim 1, characterized in that, A sensor selection decision tree is constructed, using specific data on test parameter requirements as training data to train the sensor selection decision tree, thereby filtering selection decision rule conditions, including: First, a training set for the sensor selection decision tree is established, including: finding all test parameter requirements for the parameter type of the sensor from the test parameter requirement library; and using these test parameter requirements as labels for the sensor selection decision tree training set. Secondly, in the test parameter requirement database, the specific data of the test parameter requirements in the previous flight test missions corresponding to each test parameter requirement are used as the training data of the sensor selection decision tree training set; with the help of expert knowledge, each test parameter requirement in the sensor selection decision tree training set is polled, and each sensor in the sensor information database is labeled as to whether it meets the test parameter requirements. The judgment result of whether it meets the requirements is used as the labeling result of each set of training data in the sensor selection decision tree training set. Finally, based on the ID3 decision tree algorithm, a sensor selection decision tree is constructed by calculating information gain. The sensor selection decision tree contains multiple test parameter requirements selected through training, and these test parameter requirements are used as selection decision rule conditions.

3. The intelligent decision-making method for flight test system configuration design based on business rules as described in claim 1, characterized in that, When determining the overall score, firstly, a decision rule table for the sensor selection decision tree is constructed, and then the expert composite weight corresponding to each selection decision rule condition in the decision rule table is solved. The individual scores corresponding to each selection decision rule condition are calculated again, and finally the comprehensive score of each alternative is obtained based on the expert composite weight and the individual scores.

4. The intelligent decision-making method for flight test system configuration design based on business rules according to claim 3, characterized in that, Construct a decision rule table for the sensor selection decision tree, including: Based on the selection decision rules contained in the sensor selection decision tree, the corresponding test parameter requirements are found from the test parameter requirement database, and the corresponding sensor attribute indicators are found from the sensor information database. Based on the correspondence between the selection decision rules, test parameter requirements and sensor attribute indicators, a judgment rule table is constructed.

5. The intelligent decision-making method for flight test system configuration design based on business rules according to claim 3, characterized in that, Solve for the expert composite weights corresponding to each selection decision rule condition in the decision rule table, including: The first step is to construct a best-worst fuzzy comparison matrix for all the selection decision rule conditions in the judgment rule table. ,include: The decision-making rules and conditions were determined through expert voting. Choose the most important decision rule conditions and the least important choice decision rule conditions ; Construct the first An expert on Fuzzy comparison vector , The element in is the first Paired comparisons by experts Compared to other choice decision rules and conditions The importance of the fuzzy weights assigned accordingly; Construct the first An expert on Fuzzy comparison vector , The element in is the first Experts compared the conditions of other choice decision rules with those of individual experts. The importance of each element is assigned a fuzzy weight; all fuzzy weights are represented in triplet form. Then the best-worst fuzzy comparison matrix ; The second step is to define the optimization function and solve the... Each expert analyzed the conditions of the decision-making rule for each choice. Assigned fuzzy weights ; The third step is to utilize the fuzzy weights of different experts. Solving for expert composite weights .

6. The intelligent decision-making method for flight test system configuration design based on business rules as described in claim 5, characterized in that, The optimization function is expressed as follows: in, Indicates the first One expert Assigning fuzzy weights, Indicates the first One expert Assigning fuzzy weights; Indicates the first One expert The assigned fuzzy weights are represented in triplet form as follows: , These represent the corresponding lower limit, modal value, and upper limit, respectively. Indicates the first Paired comparisons by experts Compared to The importance of the fuzzy weights assigned accordingly; Indicates the first Paired comparisons by experts Compared to The importance of the fuzzy weights assigned accordingly; To use the transformation function GMIR to fuzzy weights The result after converting to a concise value; To determine the number of conditions for selecting decision rules in the rule table.

7. The intelligent decision-making method for flight test system configuration design based on business rules according to claim 6, characterized in that, When solving the optimization function, variables are introduced. At the same time , , , , All of these are represented using the corresponding triplet form, thus transforming the objective of the optimization problem into minimizing variables. , using variables Indicates the first One expert and The ratio of the assigned fuzzy weights to The maximum absolute distance, and the first An expert became and The ratio of the assigned fuzzy weights to The maximum absolute distance is calculated to simplify the solution process.

8. The intelligent decision-making method for flight test system configuration design based on business rules according to claim 5, characterized in that, Using fuzzy weights from different experts Solving for expert composite weights Specifically: With all M A simplified value for assigning fuzzy weights by an expert. The average value is the condition of the selection decision rule. Expert Synthesis Weight .

9. The intelligent decision-making method for flight test system configuration design based on business rules according to claim 3, characterized in that, The process involves solving for the individual scores corresponding to each selection decision rule condition, and finally obtaining the comprehensive score for each alternative based on the expert composite weights and the individual scores, including: First, based on the characteristics of each selection decision rule condition, design the corresponding sensor attribute indicators, test parameter requirements, and individual score relationship formulas; the relationship formulas are divided into two cases: one is a piecewise function relationship, and the other is a selection function relationship; Secondly, the one-way score is calculated according to the relationship, and the one-way score is averaged after being weighted by the corresponding expert synthesis weight. The average value is used as the comprehensive score.

10. A terminal device, comprising a processor, a memory, and a computer program stored in the memory; characterized in that, When the processor executes the computer program, it implements the intelligent decision-making method for the configuration design of the flight test system based on business rules as described in any one of claims 1-9.

Images