A balanced uniform design and construction method for multi-type factorial experiments

Abstract

The present application relates to a balanced uniform design and construction method for multi-type factor experiments. The method includes: a calculation algorithm for realizing a balanced uniformity criterion according to the UAV equipment test space; solving a balanced uniform design problem model, storing the initialized UAV equipment test design matrix into another UAV equipment test design matrix, constructing a perturbation UAV equipment test design matrix based on another UAV equipment test design matrix, calculating the balanced uniformity criterion of the perturbation UAV equipment test design matrix and another UAV equipment test design matrix, and accepting the perturbation UAV equipment test design matrix after comparing the sizes, until the current temperature is less than a preset temperature threshold, and then outputting the final UAV equipment test design matrix. This method can simultaneously optimize the overall UAV equipment test space filling property and the balance of the classification factor sub-design level combination.

Description

Technical Field

[0001] The present application relates to the technical field of UAV equipment test design, and in particular to a balanced uniform design and construction method for multi-type factor tests. Background Art

[0002] The current UAV equipment test problems include both qualitative (categorical or ordered) and quantitative (continuous or discrete) test factors. Qualitative factors include categorical factors and ordered factors in UAV equipment tests. Categorical factors include UAV type and mission load type, among which UAV type is a typical categorical factor. For example, it can be divided into fixed-wing UAV, multi-rotor UAV and unmanned helicopter. In the test, it is important to study the performance of different types of UAVs according to different mission requirements (such as long-distance monitoring is suitable for fixed-wing UAVs, close-range complex environment operations are suitable for multi-rotor UAVs). Mission load type is also a categorical factor. For example, it can be divided into optical camera load, infrared thermal imaging load, gas sensor load, etc. Different types of loads determine the mission purpose of the UAV. In the test, it is necessary to evaluate the impact of different mission load types on the overall performance of the UAV (such as center of gravity change, flight stability), as well as the working effect of the mission load in different mission scenarios. Ordered factors include the level of manufacturing process of UAV and the level of complexity of flight control mode. The level of manufacturing process of UAV is regarded as an ordered factor. For example, it is divided into three levels: low, medium and high. UAVs with low process levels may have poor material quality, component accuracy, and assembly process, resulting in low performance and reliability of the UAVs; UAVs with medium process levels are more balanced in all aspects; UAVs with high process levels use high-quality materials, high-precision components, and advanced assembly technology, and have better performance and reliability. In the test, the impact of manufacturing processes on UAV equipment was evaluated by comparing the differences in flight performance, mission execution capabilities, and other aspects of UAVs with different process levels. The complexity level of flight control mode is an ordered factor. From a simple manual control mode to a semi-automatic control mode, and then to a fully automatic control mode. The manual control mode requires the operator to have high operating skills and make real-time adjustments to the flight attitude and trajectory; the semi-automatic control mode can assist the operator to complete some basic flight operations, such as maintaining altitude and flying on a route; in the fully automatic control mode, the UAV can automatically complete flight tasks according to the preset program, such as automatic takeoff, cruising, and landing. In the test, the indicators such as the task completion efficiency and accuracy of the UAV under different levels of control mode complexity were studied. Quantitative factors include continuous numerical factors and discrete numerical factors in the UAV equipment test. Among them, the flight speed of the UAV is a continuous numerical factor, which can take values ​​continuously within a certain range. In the experiment, the relationship between flight speed and other performance indicators was established by changing the continuous variable of flight speed and measuring the energy consumption curve of the drone and stability at different speeds. Endurance is also a continuous numerical factor, which is related to factors such as the drone's battery capacity, power system efficiency, and flight mission mode.In the test, by adjusting factors such as battery capacity and flight conditions, data such as the mission execution range and efficiency of the drone under different flight times are obtained. For example, the relationship between the flight time and the mapping mission area is studied to determine the flight time of the drone that meets the requirements of a specific mission. The number of propellers of a drone is a discrete numerical factor, and the common ones are double propellers, three propellers, four propellers, and six propellers. Different numbers of propellers will affect the lift distribution, power balance, and maneuverability of the drone. In the test, the performance differences of drones with different numbers of propellers under the same mission (such as load-bearing flight, fast turns, etc.) are compared to determine the optimal number of propellers. The communication frequency band (GHz) is a discrete numerical factor, such as 2.4GHz, 5.8GHz, etc. In the test, the stability, transmission distance, and transmission rate of drone data transmission under different communication frequency bands are studied to select the communication frequency band that best suits the mission requirements.

[0003] Providing a scientific space-filling design method for UAV equipment test problems is of great significance for improving the efficiency of UAV equipment tests. The purpose of space-filling design is to make the selected test points fill the entire design space as evenly as possible. Existing methods mainly include two categories. The first category is the design based on Latin hypercube sampling, including sliced ​​Latin hypercube design, marginally coupled design, doubly coupled design, etc.; the other category is the optimal design based on criteria, including maximum projection criterion, uniformity criterion and minimum moment aberration criterion. However, these two types of methods have inherent defects: the design based on Latin hypercube sampling has special requirements on the test sample size, and the optimal design based on criteria cannot guarantee the orthogonality of discrete factor sub-designs; some criteria cannot even optimize discrete factor designs, and it is necessary to optimize continuous factor designs based on given discrete factor designs. The sliced ​​structure of the sliced ​​Latin hypercube requires that the sample size must be an integer multiple of the discrete factor level combination. When there are many discrete factors or many levels of discrete factors, the requirement for the experimental sample size is very high. The maximum projection criterion, uniformity criterion, and minimum distortion criterion do not consider the orthogonality of the classification factor sub-design, so the design obtained by optimizing these criteria cannot guarantee orthogonality. Summary of the invention

[0004] Based on this, it is necessary to provide a balanced uniform design and construction method for multi-type factor experiments that can improve the efficiency of UAV equipment tests and the space filling of UAV equipment tests in response to the above-mentioned technical problems.

[0005] A balanced uniform design and construction method for a multi-type factor experiment, the method comprising:

[0006] Construct the UAV equipment test space according to the UAV equipment test problem; the UAV equipment test space is a space described by the value range of all test factors;

[0007] The balanced uniformity criterion is constructed to measure the filling property of the UAV equipment test space and the balance of the classification factor level combination of a given design, and an algorithm for calculating the balanced uniformity criterion based on the UAV equipment test space is given, that is, a balanced uniform design problem model is established;

[0008] Solve the balanced uniform design problem model, initialize the UAV equipment test design matrix and set the initial temperature, store the initialized UAV equipment test design matrix in another UAV equipment test design matrix, construct a perturbation UAV equipment test design matrix based on another UAV equipment test design matrix, calculate the balanced uniformity criteria of the perturbation UAV equipment test design matrix and another UAV equipment test design matrix, and then accept the perturbation UAV equipment test design matrix after comparing the sizes, until the current temperature is less than the preset temperature threshold, output the final UAV equipment test design matrix and the balanced uniform design criterion value of the final UAV equipment test design matrix.

[0009] This application proposes a new space-filling design method. The constructed balanced uniformity criterion can simultaneously ensure the overall space filling of the UAV equipment test design and the balance of the level combination of the classification factor sub-design, and designs an algorithm for solving the balanced uniform design. The algorithm ensures that the continuous factor design in the initial design and perturbation design constructed during the solution process is always a midpoint Latin hypercube design. The above-mentioned balanced uniform design and construction method of multi-type factor tests are not constrained by factor types or sample size, and also ensure the uniformity of the continuous factor design and the orthogonality of the discrete factor design in the UAV equipment test process, providing a scientific space-filling design for the UAV equipment test problem, thereby improving the efficiency of the UAV equipment test. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 A schematic diagram of a flow chart of a balanced uniform design and construction method of a multi-type factor experiment in an embodiment;

[0011] Figure 2It is a schematic diagram of a flow chart for solving the balanced uniform design problem model in an embodiment. DETAILED DESCRIPTION

[0012] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0013] In one embodiment, Figure 1 As shown, a balanced uniform design and construction method for a multi-type factor experiment is provided, including the following steps:

[0014] Step 102, constructing a UAV equipment test space according to the UAV equipment test problem; the UAV equipment test space is a space described by the value ranges of all test factors.

[0015] Test factors include qualitative factors and quantitative factors. Qualitative factors represent controllable factors that affect the test results (response variables) in the test, and their values ​​cannot be quantified and can only be limited to a finite number of abstract levels. They can be further divided into categorical factors (there is an unordered relationship between the levels of the factor) and ordered factors (there is an ordered relationship between the levels of the factor). Qualitative factors include categorical factors and ordered factors in the UAV equipment test. The categorical factors include the type of UAV and the type of mission load, among which the type of UAV is a typical categorical factor. For example, it can be divided into fixed-wing UAVs, multi-rotor UAVs, and unmanned helicopters. In the test, it is important to study the performance of different types of UAVs according to different task requirements (such as long-distance monitoring is suitable for fixed-wing UAVs, and close-range complex environment operations are suitable for multi-rotor UAVs). The type of mission load is also a categorical factor. For example, it can be divided into optical camera load, infrared thermal imaging load, gas sensor load, etc. Different types of loads determine the mission purpose of the UAV. In the test, it is necessary to evaluate the impact of different mission load types on the overall performance of the UAV (such as center of gravity change, flight stability), as well as the working effect of the mission load in different mission scenarios. Ordered factors include the level of manufacturing process of UAVs and the level of complexity of flight control modes. The level of manufacturing process of UAVs is regarded as an ordered factor. For example, it is divided into three levels: low, medium and high. UAVs with low process levels may be poor in material quality, component precision, assembly process, etc., resulting in low performance and reliability of UAVs; UAVs with medium process levels are more balanced in all aspects; UAVs with high process levels use high-quality materials, high-precision components and advanced assembly technology, and have better performance and reliability. In the experiment, the impact of manufacturing process on UAV equipment was evaluated by comparing the differences in flight performance, mission execution capabilities and other aspects of UAVs with different process levels. The level of complexity of flight control modes is an ordered factor. From simple manual control mode to semi-automatic control mode, and then to fully automatic control mode. The manual control mode requires the operator to have high operating skills and make real-time adjustments to the flight attitude and trajectory; the semi-automatic control mode can assist the operator to complete some basic flight operations, such as maintaining altitude, route flight, etc.; in the fully automatic control mode, the UAV can automatically complete flight tasks according to the preset program, such as automatic takeoff, cruising, landing, etc. In the experiment, the indicators such as the mission completion efficiency and accuracy of the UAV were studied under different levels of control mode complexity.

[0016] Quantitative factors refer to factors that have an impact on the test results (response variables), are controllable, and can be quantified. They can be further divided into continuous quantitative factors (values ​​can be taken anywhere within a continuous interval in the test) and discrete quantitative factors (values ​​can only be taken from a limited number of values ​​that are pre-defined in the test). Quantitative factors include continuous numerical factors and discrete numerical factors in the UAV equipment test. Among them, the flight speed (m / s) of the UAV is a continuous numerical factor that can take values ​​continuously within a certain range. In the test, by changing the continuous variable of flight speed, measuring the energy consumption curve of the UAV, stability at different speeds and other indicators, the relationship between flight speed and other performance indicators is established. Endurance time (hours) is also a continuous numerical factor, which is related to factors such as the battery capacity, power system efficiency, and flight mission mode of the UAV. In the test, by adjusting factors such as battery capacity and flight conditions, data such as the mission execution range and efficiency of the UAV under different endurance times are obtained. For example, the relationship between endurance time and the area of ​​surveying and mapping tasks is studied to determine the endurance time of the UAV that meets specific mission requirements. The number of propellers of a drone is a discrete numerical factor, and common ones include two propellers, three propellers, four propellers, and six propellers. Different numbers of propellers will affect the lift distribution, power balance, and maneuverability of the drone. In the experiment, the performance differences of drones with different numbers of propellers under the same mission (such as load flight, fast turning, etc.) were compared to determine the optimal number of propellers. The communication frequency band (GHz) is a discrete numerical factor, such as 2.4GHz, 5.8GHz, etc. In the experiment, the stability, transmission distance, and transmission rate of drone data transmission under different communication frequency bands were studied to select the communication frequency band that best suits the mission requirements.

[0017] UAV equipment test space It is a space described by the value range (or level) of all test factors. In the algorithm implementation, the data format shown in the example in Table 1 can be used to describe the UAV equipment test space. Each row represents a factor, and the number of rows represents the number of factors contained in the UAV equipment test space. The first column is the factor name, the second column is the factor type, and the third column is the value range or level of the factor. The factor name is named according to the physical characteristics of the factor and described by a string; the factor type is limited to four types: continuous value, discrete value, ordered and classified, and is described by four strings "Continuous", "Discrete", "Ordinary" and "Categorical" respectively; if the factor type is continuous, the third column gives its value range in the form of an interval; if the factor type is discrete, the third column gives its finite value in the form of a set; if the factor is ordered, the third column gives all possible levels of each factor in order from small to large; if the factor type is categorical, the third column can list all its levels in any order. It is recommended to use non-negative integers to encode the levels of ordered and categorical factors. The following records the number of factors contained in the UAV equipment test space , where the number of continuous quantitative factors is , the number of discrete quantitative factors and ordered factors is , the number of classification factors is ,here , , and For simplicity, it is assumed that the front The factors are continuous quantitative factors, and the intermediate The factors are discrete quantitative or ordered factors. The factors are categorical factors.

[0018] Table 1

[0019]

[0020] Step 104, construct a balanced uniformity criterion to measure the filling property of the UAV equipment test space and the balance of the classification factor level combination of a given design, and provide an algorithm for calculating the balanced uniformity criterion based on the UAV equipment test space, that is, establish a balanced uniform design problem model.

[0021] Balanced uniform design is an optimal design based on criteria. A UAV equipment test design matrix is a Order Matrix , each row represents a test point, and each column represents the value of a factor in each test point. The value range is determined by the UAV equipment test space. In actual UAV equipment tests, the design matrix will be designed based on specific test objectives, research focus, and operational resources. It contains multiple test factors, such as flight altitude, environmental conditions, communication frequency bands, etc., and the level of each factor is divided and combined in a more detailed manner to comprehensively and systematically evaluate the performance and characteristics of UAV equipment.

[0022] Balance means that for different level combinations of discrete factors, each combination has a relatively fair chance of appearing in the experiment, which can avoid the situation where some combinations are over-tested or some combinations are under-tested. Uniformity is mainly aimed at filling the test space, which means that the test points are distributed as evenly as possible in the entire test space to fully explore all areas of the test space.

[0023] In this application, a UAV equipment test design matrix is ​​given And strength , equilibrium uniformity criterion is defined as follows:

[0024] ;

[0025] The meaning of each letter is as follows.

[0026] (1) Continuous quantitative factor part:

[0027] ;

[0028] Note, here The corresponding factor is a continuous factor. Representation Matrix No. Line The normalized values ​​of the column elements; the normalization method for continuous factors is as follows

[0029] ;

[0030] Representation Matrix No. Line Column elements, and They represent the first The lower and upper bounds of the range of values ​​of a continuous quantitative factor.

[0031] (2) Discrete quantitative factors and ordered factors:

[0032] ;

[0033] here Indicates factors (note that here corresponding to discrete quantitative or categorical factors) Levels or matrices No. Line The normalized values ​​of the column elements (determined by the range of the sum), Indicates The number of levels of a factor; the normalization method of discrete quantitative factors is

[0034] ;

[0035] in, Indicates The factor Levels or matrices No. Line The value of the column element,

[0036] ;

[0037] Respectively represent the maximum and minimum values ​​of the factor in the sample space,

[0038] ;

[0039] represents half of the minimum interval between all possible values ​​of the factor in the sample space; the normalization method of ordered factors is to first use the digital The levels are coded and then

[0040] normalized in the same way.

[0041] (3) Classification factor part:

[0042] ;

[0043] in, and Respectively represent matrices No. Line Column and Line Column elements, Indicates The number of levels of the factor (note that here corresponds to the categorical factor), constant and Need to meet , recommended , .

[0044] (4) Orthogonality of classification factors: Indicates from Select from the categorical factors The first Among the possible results, the design The number of occurrences of the most frequent level combination in ; Indicates from Categorical factor selection The first Among the possible results, the design The number of level combinations that occur least often in ; therefore, if the categorical factor subdesign constitutes a strength An orthogonal array of , otherwise ideally Should be equal to 1.

[0045] When optimizing the test space filling, the importance and correlation of variables also need to be considered. For UAV equipment tests, some variables may have a greater impact on the test results. For example, the correlation between flight speed and flight stability may be high, while there is also a close connection between fuselage weight and battery life.

[0046] Therefore, it is necessary to optimize the filling property of the UAV equipment test space and the balance of the discrete factor design level combination of the UAV equipment test space. This application optimizes the filling property of the UAV equipment test space and the balance of the discrete factor design level combination of the UAV equipment test space according to the balanced uniformity criterion, and obtains the balanced uniform design problem model as follows:

[0047] .

[0048] Step 106, solve the balanced uniform design problem model, initialize the UAV equipment test design matrix and set the initial temperature, store the initialized UAV equipment test design matrix in another UAV equipment test design matrix, construct a perturbation UAV equipment test design matrix based on the other UAV equipment test design matrix, calculate the balanced uniformity criteria of the perturbation UAV equipment test design matrix and the other UAV equipment test design matrix, and then accept the perturbation UAV equipment test design matrix after comparing the sizes, until the current temperature is less than the preset temperature threshold, and then output the final UAV equipment test design matrix and its balanced uniform design criterion value.

[0049] The balanced uniform design problem model is a high-dimensional optimization problem, and its optimization variable is the UAV equipment test design matrix of elements. Sample size is a positive integer, generally at least 10 times the number of factors contained in the UAV equipment test space; intensity is a positive integer whose value does not exceed the number of classification factors in the UAV equipment test space , The recommended value of is 2. This application designs a process for solving the balanced uniform design problem model, such as Figure 2 As shown, the specific steps are as follows:

[0050] Initialization includes constructing the initial design and setting the initial temperature The initial temperature is directly set to 10,000 degrees, and the initial UAV equipment test design matrix The construction method is to construct its first List:

[0051] If The columns correspond to continuous factors. First generate A random permutation of , then let

[0052] ;

[0053] If The columns correspond to discrete factors, with To represent its level number, first let , then No. Front of column elements are set to this factor level Repeat, then Elements from It is obtained by simple random sampling without replacement in the levels.

[0054] The initial UAV equipment test design matrix constructed according to the above method can ensure that the continuous factor sub-design is a mid-point Latin hypercube design, and the one-dimensional projection of the discrete factor sub-design has the greatest balance.

[0055] The counter Set to 1 and design Store into another matrix middle.

[0056] exist The perturbation design matrix is ​​constructed based on The construction method is as follows:

[0057] First, , then from of Select one of the columns, assuming the selected column is List;

[0058] If If the column corresponds to a continuous factor, two elements are randomly selected from the column and swapped;

[0059] If If the column corresponds to a discrete factor, then a random element is selected from the column ,if The number of times it appears in this column is , then randomly select two more elements from the column and , and rotate these three elements; if The number of times it appears in this column is , then randomly select another two elements with a probability of 0.5 and rotate the three elements, and then randomly select the other two elements with a probability of 0.5. Replace with another randomly selected occurrence of The level of the next.

[0060] calculate and The equilibrium uniformity criterion and , and compare their sizes; if , then accept the disturbance order Otherwise The probability of accepting the perturbation order .

[0061] Compare and ,if , then let as well as Otherwise, .

[0062] if , then re- Constructing disturbance design based on .

[0063] make ,if Then reset the counter Set to 1 and design Store into another matrix middle.

[0064] The final solution is an approximate solution for the balanced uniform design , and its balanced uniform design criterion value .

[0065] This application proposes a new space-filling design method. The balanced uniformity criterion of the construction can simultaneously ensure the overall space filling of the UAV equipment test and the balance of the level combination of the classification factor sub-design, and designs an algorithm for solving the balanced uniform design. The algorithm ensures that the continuous factor design in the initial design and perturbation design constructed during the solution process is always a midpoint Latin hypercube design. The above-mentioned balanced uniform design and construction method of a multi-type factor experiment is not restricted by the factor type and the sample size, and also ensures the uniformity of the continuous factor design and the orthogonality of the classification factor design.

[0066] In one embodiment, the equilibrium uniformity criterion is

[0067] ;

[0068] in, represents the UAV equipment test design matrix, Indicates strength, , Calculated from the continuous factor part of the UAV equipment test design matrix, , , It is calculated from the discrete quantitative factors and ordered factors of the UAV equipment test design matrix. , , Calculated from the classification factor part of the UAV equipment test design matrix, , Calculated from the discrete factor part of the UAV equipment test design matrix, represents the number of discrete quantitative and ordered factors, represents the number of classification factors, is the sample size, i.e. the number of test points.

[0069] In one embodiment, the three items calculated from the continuous factor part of the UAV equipment test design matrix are: the first item is ; The second item is ; The third item is ;in, It indicates that it corresponds to a continuous quantitative factor. Represents the UAV equipment test design matrix No. Line The normalized values ​​of the column elements.

[0070] In one embodiment, the three items calculated from the discrete quantitative factors and ordered factors of the UAV equipment test design matrix are: the first item is ,in and Respectively represent The factor and The horizontally normalized values, Indicates The number of levels of the factor, here Corresponds to discrete quantitative or categorical factors, binary functions Defined as ; The second item is ,in Represents the UAV equipment test design matrix No. Line The normalized value of the column elements, Indicates The factor The value after horizontal normalization; the third term is ,in and Respectively represent the UAV equipment test design matrix No. Line Column elements and Line The normalized values ​​of the column elements.

[0071] In one embodiment, the three items calculated from the classification factor part of the UAV equipment test design matrix are: the first item is ; The second item is ; The third item is ;in, and Respectively represent the UAV equipment test design matrix No. Line Column and Line Column elements, Indicates The number of levels of the factor, Corresponding to the categorical factor, the constant and Need to meet , recommended , .

[0072] In one embodiment, Indicates from Select from the categorical factors The factor Among the possible results, the UAV equipment test design matrix The number of occurrences of the most frequent level combination in ; Indicates from Discrete Factor Selection The first Among the possible results, the UAV equipment test design matrix The number of occurrences of the level combination with the least occurrence in .

[0073] In one embodiment, a balanced uniformity criterion is constructed to measure the filling property of the UAV equipment test space and the balance of the classification factor level combination of a given design, and an algorithm for calculating the balanced uniformity is implemented according to the UAV equipment test space, and the balanced uniform design problem model is obtained as follows:

[0074] .

[0075] In one embodiment, initializing the UAV equipment test design matrix includes:

[0076] Construct the first part of the UAV equipment test design matrix Column, if The columns correspond to continuous factors. First generate A random permutation of , then let

[0077] ;

[0078] If The columns correspond to discrete factors, with To represent its level number, first let , and then the first Front of column elements are set to this factor level Repeat, then Elements from It is obtained by simple random sampling without replacement in the levels.

[0079] In one embodiment, a disturbance UAV equipment test design matrix is ​​constructed based on another UAV equipment test design matrix, including: making the disturbance UAV equipment test design matrix ,in Design a matrix for another UAV equipment test, and then of Select one of the columns, assuming the selected column is Column; if If the column corresponds to a continuous factor, two elements are randomly selected from the column and swapped; if If the column corresponds to a discrete factor, then a random element is selected from the column ,if The number of times it appears in this column is , then randomly select two more elements from the column and , and rotate these three elements; if The number of times it appears in this column is , then randomly select another two elements with a probability of 0.5 and rotate the three elements, and then randomly select the other two elements with a probability of 0.5. Replace with another randomly selected occurrence of The level of the next.

[0080] In one embodiment, the balanced uniformity criterion of the disturbance UAV equipment test design matrix and another UAV equipment test design matrix is ​​calculated and the disturbance UAV equipment test design matrix is ​​accepted after the size is compared, including:

[0081] Computational disturbance UAV equipment test design matrix And another UAV equipment test design matrix The equilibrium uniformity criterion and , and compare their sizes; if , then accept the disturbance order Otherwise The probability of accepting the perturbation order .

[0082] It should be understood that although Figure 1 The steps in the flowchart are shown in sequence as indicated by the arrows, but these steps are not necessarily executed in the order indicated by the arrows. Unless otherwise specified in this document, there is no strict order restriction for the execution of these steps, and these steps can be executed in other orders. Moreover, Figure 1 At least part of the steps may include multiple sub-steps or multiple stages. These sub-steps or stages are not necessarily executed at the same time, but can be executed at different times. The execution order of these sub-steps or stages is not necessarily sequential, but can be executed in turn or alternately with other steps or at least part of the sub-steps or stages of other steps.

[0083] Those of ordinary skill in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be implemented by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media used in the embodiments provided in this application may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in many forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM), etc.

[0084] The technical features of the above embodiments may be arbitrarily combined. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0085] The above-mentioned embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the invention patent. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the patent of the present application shall be subject to the attached claims.

Claims

1. A balanced uniform design and construction method for multi-type factor experiments, characterized in that: The method comprises: Constructing a UAV equipment test space according to the UAV equipment test problem; the UAV equipment test space is a space described by the value ranges of all test factors; Construct a balanced uniformity criterion to measure the filling property of the UAV equipment test space and the balance of the classification factor level combination of a given design, and provide an algorithm for calculating the balanced uniformity criterion based on the UAV equipment test space, that is, establish a balanced uniform design problem model; Solve the balanced uniform design problem model, initialize the UAV equipment test design matrix and set the initial temperature, store the initialized UAV equipment test design matrix in another UAV equipment test design matrix, construct a perturbation UAV equipment test design matrix based on the other UAV equipment test design matrix, calculate the balanced uniformity criteria of the perturbation UAV equipment test design matrix and the other UAV equipment test design matrix, and accept the perturbation UAV equipment test design matrix after comparing the sizes, until the current temperature is less than a preset temperature threshold, output the final UAV equipment test design matrix and the balanced uniform design criterion value of the final UAV equipment test design matrix; The equilibrium uniformity criterion is ； in, represents the UAV equipment test design matrix, Indicates strength, , , Calculated from the continuous factor part of the UAV equipment test design matrix, , , It is calculated from the discrete quantitative factors and ordered factors of the UAV equipment test design matrix. , , Calculated from the classification factor part of the UAV equipment test design matrix, , Calculated from the classification factor part of the UAV equipment test design matrix, represents the number of classification factors, is the sample size, i.e. the number of test points.

2. The method according to claim 1, characterized in that The three items calculated from the continuous factor part of the UAV equipment test design matrix are: the first item is ; The second item is ; The third item is ;in, Represents the UAV equipment test design matrix No. Line The normalized value of the column elements, here The corresponding factor is a continuous quantitative factor.

3. The method according to claim 1, characterized in that: The three items calculated from the discrete quantitative factors and ordered factors of the UAV equipment test design matrix are: ,in and Respectively represent The factor and The horizontally normalized values, Indicates The number of levels of the factor, here Corresponding to discrete quantitative or categorical factors, the binary function definition for ; The second item is ,in Represents the UAV equipment test design matrix No. Line The normalized value of the column elements, Indicates The factor The value after horizontal normalization; the third term is ,in and Respectively represent the UAV equipment test design matrix No. Row Column Elements and Line The normalized values ​​of the column elements.

4. The method according to claim 1, characterized in that The three items calculated from the classification factor part of the UAV equipment test design matrix are: the first item is ; The second item is ; The third item is ;in, and Respectively represent the UAV equipment test design matrix No. Line Column and Line Column elements, Indicates The number of levels of the factor, here corresponds to the categorical factor; the constant and Need to meet This application recommends .

5. The method according to claim 1, characterized in that Said Indicates from Select from the categorical factors The factor Among the possible results, the UAV equipment test design matrix The number of occurrences of the most frequent level combination in ; Indicates from Categorical factor selection The first Among the possible results, the UAV equipment test design matrix The number of occurrences of the level combination with the least occurrence in .

6. The method according to claim 1, characterized in that According to the algorithm for calculating the balanced uniformity of the UAV equipment test space, the balanced uniformity design problem model is obtained as follows: 。 7. The method according to claim 1, characterized in that Initialize the UAV equipment test design matrix, including: Construct the first part of the UAV equipment test design matrix Column, if The columns correspond to continuous factors. First generate A random permutation of , then let ； If The columns correspond to discrete factors, with To represent its level number, first let , and then the first Front of column elements are set to this factor level Repeat, then Elements from It is obtained by simple random sampling without replacement in the levels.

8. The method according to claim 1, characterized in that The disturbance UAV equipment test design matrix is ​​constructed based on another UAV equipment test design matrix, including: let the disturbance UAV equipment test design matrix ,in Design a matrix for another UAV equipment test, and then of Select one of the columns, assuming the selected column is Column; if If the column corresponds to a continuous factor, two elements are randomly selected from the column and swapped; if If the column corresponds to a discrete factor, then a random element is selected from the column ,if The number of times it appears in this column is , then randomly select two more elements from the column and , and rotate these three elements; if The number of times it appears in this column is , then randomly select another two elements with a probability of 0.5 and rotate the three elements, and then randomly select the other two elements with a probability of 0.

5. Replace with another randomly selected occurrence of The level of the next.

9. The method according to claim 1, characterized in that: Calculating the equilibrium uniformity criterion of the disturbance UAV equipment test design matrix and another UAV equipment test design matrix and comparing the sizes, and then accepting the disturbance UAV equipment test design matrix, including: Computational disturbance UAV equipment test design matrix And another UAV equipment test design matrix The equilibrium uniformity criterion and , and compare their sizes; if , then accept the disturbance order Otherwise The probability of accepting the perturbation order .

Images