A dynamic formation adjustment method for unmanned ship formation considering switching cost

Abstract

The present invention discloses a method for adjusting the dynamic formation of an unmanned ship formation considering the switching cost, S1: constructing an unmanned ship formation energy consumption calculation model, the unmanned ship formation energy consumption calculation model is used to calculate the total energy consumption of the formation; S2: constructing a formation energy consumption historical data matrix and a formation energy consumption current data matrix; S3: constructing a dynamic objective function, and obtaining a current dynamic data matrix based on the dynamic objective function and the formation energy consumption current data matrix; constructing a formation energy consumption future data matrix, thereby obtaining a formation energy consumption time domain database; S4: determining a reasonable local optimal decision variable based on a time domain robust optimization algorithm and a dynamic objective function; S5: optimizing the reasonable local optimal decision variable using a wild dog optimization algorithm, and finally obtaining an optimal spacing configuration sequence; the constructed unmanned ship formation formation is an unmanned ship energy-saving formation formation. The present invention not only provides a low-cost scheduling formation switching decision for the unmanned ship formation to continue sailing in the waves, but also ensures that the adjusted formation still has an energy-saving effect, realizes dynamic energy saving of the unmanned ship formation in long-distance navigation, and plays a key role in promoting the development of green shipping for ships.

Description

Technical Field

[0001] The present invention relates to the field of unmanned ships and multi-agent navigation technology, and in particular to a method for adjusting the dynamic formation of an unmanned ship fleet taking switching costs into consideration. Background Art

[0002] The formation of unmanned surface ships has gradually become an important force in ocean exploration and development. When faced with complex marine operations, the formation of unmanned ships can fully utilize the advantages of cluster collaboration and effectively make up for the performance deficiencies of a single ship, thus ensuring the smooth completion of the mission.

[0003] In order to overcome the influence of wind and waves during navigation, the unmanned ship formation will adopt a formation adjustment method to make the formation sail in an energy-saving formation layout, thereby improving the overall endurance of the formation. During the formation adjustment process, a large change in the position of the members will not only bring certain difficulties to the operation of the unmanned ship, but also very likely cause the main engine and propeller to operate excessively, resulting in increased battery consumption. Even if the adjusted formation achieves energy-saving effects, the long-distance dispatching causes the unmanned ship formation to consume a lot of electricity. This energy loss far exceeds the energy-saving benefits obtained by adopting the energy-saving formation, which violates the purpose of green navigation of the unmanned ship formation. Summary of the invention

[0004] The present invention provides a method for dynamic formation adjustment of an unmanned ship formation taking switching costs into consideration, so as to overcome the technical problem that when the unmanned ship formation is adjusting its formation, the energy consumed by switching the formation may exceed the energy-saving benefits brought by switching the formation due to failure to consider the switching costs, which does not meet the purpose of energy saving.

[0005] In order to achieve the above object, the technical solution of the present invention is:

[0006] S1: Obtaining a number of total formation resistances, and constructing an unmanned ship formation energy consumption calculation model based on the total formation resistances, wherein the unmanned ship formation energy consumption calculation model is used to calculate the total formation energy consumption;

[0007] S2: constructing a formation energy consumption historical data matrix and a formation energy consumption current data matrix based on the formation total resistance and the unmanned ship formation energy consumption calculation model;

[0008] S3: construct a dynamic objective function based on the unmanned ship formation energy consumption calculation model, and obtain a current dynamic data matrix based on the dynamic objective function and the current data matrix of formation energy consumption; construct a future data matrix of formation energy consumption based on the historical data matrix of formation energy consumption and the current dynamic data matrix, so as to obtain all spacing configurations and energy consumption values ​​under the whole voyage, that is, a formation energy consumption time domain database;

[0009] S4: Preliminarily determine reasonable local optimal decision variables and local minimum switching costs corresponding to the reasonable local optimal decision variables from the formation energy consumption time domain database based on the time domain robust optimization algorithm and the dynamic objective function, wherein the decision variables are the spacing configuration sequence of the unmanned ship formation;

[0010] S5: The wild dog optimization algorithm is used to optimize the reasonable local optimal decision variables, and finally the global optimal decision variables are obtained, that is, the optimal spacing configuration sequence that takes into account both minimum energy consumption and minimum switching cost; the unmanned ship formation constructed based on the optimal spacing configuration sequence is the unmanned ship energy-saving formation.

[0011] Furthermore, in S1, the process of obtaining a plurality of formation total resistances and constructing an unmanned ship formation energy consumption calculation model based on the formation total resistances is as follows:

[0012] S11: Assume an unmanned ship formation consisting of N unmanned ships. Each formation is determined by N-1 sets of spacing configurations. The corresponding number of spacing configurations is 2(N-1). Each formation corresponds to a spacing configuration sequence, recorded as (d1, d2, d3, ···d i ,···,d N-1 ,s1,s2,s3,···s i ,···,s N-1 ), where d i is the i-th longitudinal offset before the environment switch, s i is the i-th lateral deviation before the environment switch;

[0013] S12: Set the lateral deviation and longitudinal offset of each unmanned ship in the unmanned ship formation composed of N unmanned ships to not exceed the maximum communication radius L of the unmanned ship, set the increment of the spacing configuration to a, then the number of spacing configuration sequences is And M is an integer, that is, the number of unmanned ship formations is M;

[0014] The CFD numerical simulation technology is used to simulate the formation of M unmanned ships and the total resistance of the M formations is obtained.

[0015] For any unmanned ship formation among the M unmanned ship formations, the resistance of all unmanned ships in the unmanned ship formation when a single ship is self-propelled is added to obtain the total self-propelled resistance, and the total formation resistance is divided by the total self-propelled resistance to achieve dimensionless processing, and the result obtained is the overall energy consumption of the formation in the current formation;

[0016] S13: Taking M spacing configuration sequences as input variables and the overall energy consumption of the formation corresponding to the M spacing configuration sequences as output variables, a response surface approximate model is constructed for the total formation resistance corresponding to the M unmanned ship formations. The energy consumption calculation model of the unmanned ship formation is obtained by using a high-order response surface fitting method. Its mathematical form is as follows:

[0017] En=α1d1+α2d2+,···,+α N-1 d N-1 +β1s1+β2s2+,···,+β N-1 s N-1

[0018] In the formula, α1, α2, …, α N-1 and β1,β2,…,β N-1 are the regression coefficients of the i-th longitudinal offset and the i-th lateral deviation respectively; En is the total energy consumption of the formation, referred to as the energy consumption value.

[0019] Further, in S2, the process of constructing a formation energy consumption historical data matrix and a formation energy consumption current data matrix based on the formation total resistance and the unmanned ship formation energy consumption calculation model is as follows:

[0020] S21: Sort the total resistances of the M formations according to the formation numbers, take the total resistances of the first M-1 formations as the historical data of formation energy consumption, take the spacing configuration sequence corresponding to the total resistances of the first M-1 formations as the unit row vector, arrange them from top to bottom in the order of the formation numbers, and take the energy consumption value corresponding to each spacing configuration sequence as the last column, so as to construct a formation energy consumption historical data matrix;

[0021] S22: The Mth formation total resistance among the M formation total resistances is used as the current data of formation energy consumption, the spacing configuration sequence corresponding to the Mth total resistance is used as the unit row vector, and the corresponding energy consumption value is used as the last column to construct the current data matrix of formation energy consumption.

[0022] Further, in S3, a dynamic objective function is constructed based on the unmanned ship formation energy consumption calculation model, and a current dynamic data matrix is ​​obtained based on the dynamic objective function and the current data matrix of the formation energy consumption; a future data matrix of the formation energy consumption is constructed based on the historical data matrix of the formation energy consumption and the current dynamic data matrix, so as to obtain all spacing configurations and energy consumption values ​​under the whole voyage, that is, the process of the formation energy consumption time domain database is:

[0023] S31: constructing a formation dynamic environment using an improved moving peak function, wherein the improved moving peak function is expressed as:

[0024]

[0025] Where f(x,α(t)) represents the moving peak function constructed in the form of multi-dimensional and multi-peak peak function;

[0026] H i (t), W i (t) and C i (t) represents the peak height, peak width and center position of the i-th moving peak function at time t; m is the number of peaks; x represents the spacing configuration sequence;

[0027] S32: Set the movement form of the formation dynamic environment, namely:

[0028] Assume that the peak height, peak width and center position of the moving peak function at time t+1 are in the following form:

[0029] H i (t+1)=H i (t)+height_severity i Rand(0,1)

[0030] W i (t+1)=W i (t)+width_severity i Rand(0,1)

[0031] C i (t+1)=C i (t)+ν i (t+1)

[0032] In the formula, H i (t+1), W i (t+1) and C i (t+1) represents the peak height, peak width and center position of the i-th moving peak function at time t+1; height_severity i and width_severity i Respectively represent the degree of change in peak height and peak width of the i-th peak; ν i represents a moving vector; Rand(0,1) represents a random number generated between [0,1];

[0033] The moving direction and moving distance of the center position of the moving peak function and the degree of change of the peak height and peak width are set, so as to realize the dynamic change of the moving peak function;

[0034] S33: construct dynamic objective function;

[0035] The energy consumption calculation model of the unmanned ship formation is multiplied by the improved moving peak function to construct a dynamic function of the formation energy consumption, that is, a dynamic objective function. The form of the dynamic objective function is as follows:

[0036] E(x,α(t))=f(x,α(t))·En Δ

[0037] Where E(x,α(t)) represents the dynamic objective function, f(x,α(t)) is the improved moving peak function, En is the energy consumption calculation model of the unmanned ship formation, and Δ is the time-varying parameter;

[0038] S34: Substitute the current data matrix of formation energy consumption into the dynamic objective function to obtain a current dynamic data matrix;

[0039] S35: Taking the formation energy consumption history data matrix and the spacing configuration sequence in the current dynamic data matrix as input variables, using a predictor model to predict the spacing configuration sequence at a future moment and output it. The predictor model is as follows:

[0040]

[0041] Where, X d ' s The sequence of spacing configurations for future moments predicted by the predictor model; The spacing configuration sequence reserved for the environment at the previous moment; λ is the predictor model coefficient; X ds is the spacing configuration sequence in the formation energy consumption historical data matrix and the formation energy consumption current data matrix input into the predictor model; P is the training parameter of the predictor model; Δt is the number of training environments;

[0042] Among them, the training parameter P of the predictor model is iteratively trained according to the preset environmental gradient in the current environment, and then the trained P is fed back to the predictor model to change X d ' s The result; then the current moment environment is switched to the next moment environment, and P performs the next self-training; this is repeated until the number of environment switches reaches the preset upper limit of the number of environments, the training parameters P of the predictor model stop self-training, the predictor model prediction process stops, and all X are output d ' s ;

[0043] The predicted spacing configuration sequences of all future moments are taken as unit row vectors, and arranged in order from top to bottom according to the order of environment numbers. The energy consumption value corresponding to each spacing configuration sequence is calculated according to the energy consumption calculation model of the unmanned ship formation, and the corresponding energy consumption value is taken as the last column in turn, thereby constructing the future data matrix of the formation energy consumption;

[0044] S36: Arrange the formation energy consumption historical data matrix, the current dynamic data matrix and the future data matrix in chronological order, so as to obtain a formation energy consumption time domain database.

[0045] Further, in S4, based on the time domain robust optimization algorithm and the dynamic objective function, a process of preliminarily determining a reasonable local optimal decision variable from the formation energy consumption time domain database and a local minimum switching cost corresponding to the reasonable local optimal decision variable is as follows:

[0046] S41: Design a formula for calculating the cost of switching formations

[0047] The switching cost is measured by the Euclidean distance between two adjacent groups of decision variables in the formation energy consumption time domain database, thereby characterizing the degree of change of decision variables in different environments. The decision variables are the spacing configuration sequence. The calculation formula of the formation switching cost is as follows:

[0048]

[0049] Where, d i and represents the i-th longitudinal offset before and after the environment switch, s i and represents the i-th lateral deviation before and after the environment switch, || ||2 represents the second norm, N is the number of formation members, M is the number of formations, K is the total number of environments, and Ec is the formation switching cost;

[0050] S42: Design execution environment switching operation judgment process:

[0051] The execution environment switching operation judgment process is as follows: determine whether to execute the environment switching operation according to the dynamic objective function value corresponding to the decision variable at time t, that is, if E(x, α(t)) ≥ 0.9E(x0, α(t0)), it is determined that the decision variable at time t does not meet the current environment requirements, then execute the environment switching operation process, otherwise the execution environment switching operation is terminated, and the time domain robust optimization algorithm outputs the decision variable at the current moment, which is the initial local optimal decision variable; wherein x0 represents the decision variable under the initial environment, t0 represents the initial moment; E represents the dynamic function of formation energy consumption;

[0052] The process of the environment switching operation is as follows: updating the environment conditions at time t to the environment conditions at time t+1, updating the environment parameters to the moving peak function parameters, and adjusting the values ​​of the environment parameters according to the dynamic change form of the moving peak function; obtaining the decision variables and their energy consumption values ​​at time t+1 from the future data matrix of the formation energy consumption, and if all the future data of the formation energy consumption have been called, executing S43, otherwise calculating the moving peak function value of the decision variable at time t+1 according to the moving peak function, and then obtaining the dynamic objective function value of the decision variable at time t+1 in the new environment, and replacing the dynamic objective function value at time t with the dynamic objective function value at time t+1, and updating it to E(x, α(t)), and replacing the decision variable at time t with the decision variable at time t+1;

[0053] S43: The environment switching operation is terminated, and the time-domain robust optimization algorithm outputs the decision variables at the last moment, and uses the decision variables at the last moment as the initial local optimal decision variables;

[0054] S44: Check the rationality of the initial local optimal decision variables

[0055] The initial local optimal decision variable is tested according to the number of environment switching, and the testing rule is: if the number of environment switching corresponding to the initial local optimal decision variable is 1 or the set maximum number of environments, and the set maximum number of environments is not 1, then the initial local optimal decision variable is determined to be unreasonable. At this time, the unreasonable initial local optimal decision variable is eliminated and S45 is executed; otherwise, a reasonable local optimal decision variable is obtained;

[0056] S45: changing the spacing configuration sequence and the corresponding energy consumption value in the current dynamic data matrix to the spacing configuration sequence and the corresponding energy consumption value at the previous moment, that is, the spacing configuration sequence and the corresponding energy consumption value in the formation energy consumption history data matrix, and then repeatedly executing S41-S44 until a reasonable local optimal decision variable is obtained or the formation energy consumption history data is called up, then terminating the calculation;

[0057] S46: According to the formation switching cost formula, the switching cost between the reasonable local optimal decision variable and the decision variable at time t is calculated, which is the local minimum switching cost.

[0058] Furthermore, in S5, the wild dog optimization algorithm is used to optimize the reasonable local optimal decision variables, and finally the global optimal decision variables are obtained, that is, the process of obtaining the optimal spacing configuration sequence that takes into account the minimum energy consumption and the minimum switching cost is:

[0059] S51: Calculate the fitness value;

[0060] The wild dog optimization algorithm is used to randomly generate several decision variables of the same dimension according to the sequence dimension of the reasonable local optimal decision variable, which is called individuals. All individuals are arranged from top to bottom in the order of generation to form a population matrix, which is called the initial population. Among them, the reasonable local optimal decision variable is placed in the first row of the initial population as the current optimal individual. The energy consumption value of the current optimal individual is calculated according to the formation energy consumption calculation model. The energy consumption value of the current optimal individual is added to the local minimum switching cost and the inverse is taken to obtain the current fitness value. The calculation formula is as follows:

[0061]

[0062] Where ω1 and ω2 are weight coefficients obtained after normalization, and their sum is 1; En is the total energy consumption of the formation, that is, the energy consumption value; Ec is the formation switching cost;

[0063] S52: The process of evolving the reasonable local optimal decision variables using the wild dog optimization algorithm is as follows:

[0064] S521: Set the evolution principle as the principle of minimizing both the total energy consumption of the formation and the cost of switching the formation, that is, the principle of maximizing the fitness value;

[0065] S522: Execute the first round of evolution based on the wild dog optimization algorithm, perform population evolution operation on the first row of individuals in the initial population, obtain new individuals after evolution, that is, the spacing configuration sequence of the unmanned ship formation, calculate the formation switching cost of the new individuals and the original individuals through the formation switching cost calculation formula, calculate the energy consumption value of the new individuals through the formation energy consumption calculation model, and then obtain the fitness value of the new individuals;

[0066] S523: Determine whether the fitness value of the new individual is greater than the fitness value of the original individual before evolution. If so, determine that the new individual is superior, and replace the current optimal individual with the new individual, and replace the current fitness value with the fitness value of the new individual; otherwise, determine that the original individual is superior, and retain the current optimal individual and the current fitness value;

[0067] S524: According to the step content of S522, the population evolution operation is performed on the remaining rows of individuals in the initial population in turn, and then S523 is repeated to perform the judgment operation until all individuals in the initial population are evolved. Finally, the current optimal individual obtained by judgment is the global optimal decision variable, and the obtained global optimal decision variable is the optimal spacing configuration sequence that takes into account both minimum energy consumption and minimum switching cost.

[0068] Beneficial effects: The present invention constructs an unmanned ship formation energy consumption calculation model, and constructs a formation energy consumption time domain database based on the unmanned ship formation energy consumption calculation model, proposes the concept of switching cost based on Euclidean distance, characterizes the scheduling cost of switching from the original formation to the new formation, and designs a time domain robust optimization algorithm for evaluating the switching cost, and determines reasonable local optimal decision variables based on the formation energy consumption time domain database and the dynamic objective function; based on the wild dog optimization algorithm, the minimum energy consumption and the minimum switching cost are used as the optimization objectives of the spacing configuration to solve the global optimal decision variables, thereby determining the energy-saving formation of the unmanned ship. The unmanned ship formation dynamic formation adjustment method proposed by the present invention not only provides a low-cost scheduling formation switching decision for the unmanned ship formation to continue sailing in the waves, but also ensures that the adjusted formation still has an energy-saving effect, realizes dynamic energy saving of the unmanned ship formation in long-distance navigation, and plays a key role in promoting the development of green shipping. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0070] Figure 1 It is a flow chart of a method for adjusting the dynamic formation of an unmanned ship fleet taking into account the switching cost in the present invention;

[0071] Figure 2 It is a schematic diagram of the formation reconstruction of an unmanned ship fleet based on the spacing configuration transformation in an embodiment of the present invention;

[0072] Figure 3 Schematic diagram of long-distance dispatching of an unmanned ship fleet in an embodiment of the present invention;

[0073] Figure 4 Schematic diagram of close-range dispatching of an unmanned ship formation in an embodiment of the present invention;

[0074] Figure 5 Schematic diagram of an initial environment movement peak and a termination environment movement peak in an embodiment of the present invention;

[0075] Figure 6 It is a framework diagram of a time domain robust optimization algorithm considering switching cost in an embodiment of the present invention;

[0076] Figure 7 This is a result diagram of energy consumption values ​​of the energy-saving formation after dynamic formation adjustment in an embodiment of the present invention. DETAILED DESCRIPTION

[0077] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0078] This embodiment provides a method for adjusting the dynamic formation of an unmanned ship fleet taking into account the switching cost. Figure 1 As shown, the specific steps include:

[0079] S1: Obtaining a number of total formation resistances, and constructing an unmanned ship formation energy consumption calculation model based on the total formation resistances, wherein the unmanned ship formation energy consumption calculation model is used to calculate the total formation energy consumption;

[0080] In a specific embodiment, in S1, the process of obtaining a plurality of formation total resistances and constructing an unmanned ship formation energy consumption calculation model based on the formation total resistances is as follows:

[0081] S11: defining the spacing configuration of the unmanned ship formation according to the relative distance between each two adjacent unmanned ships in the unmanned ship formation on the horizontal plane;

[0082] Among them, the relative distance between two adjacent unmanned ships on the horizontal plane includes the lateral spacing, i.e., lateral deviation, and the longitudinal spacing, i.e., longitudinal offset. Since the relative distance between every two unmanned ships on the horizontal plane is determined by a lateral deviation and a longitudinal offset, the lateral deviation and longitudinal offset between every two adjacent unmanned ships in the unmanned ship formation are defined as a set of spacing configurations; then an unmanned ship formation consisting of N unmanned ships is set, and each formation is determined by N-1 sets of spacing configurations, and the corresponding number of spacing configurations is 2(N-1), and each formation corresponds to a spacing configuration sequence, recorded as (d1, d2, d3, ···d i ,···,d N-1 ,s1,s2,s3,···s i ,···,s N-1 ), where d i is the i-th longitudinal offset before the environment switch, s i is the i-th lateral deviation before the environment switch; any d i or i Any change will change the element value in the spacing configuration sequence, thereby generating a new spacing configuration sequence, which in the application scenario indicates that the unmanned ship formation has undergone a formation adjustment, such as Figure 2 As shown;

[0083] S12: Set the lateral deviation and longitudinal offset of each unmanned ship in the unmanned ship formation composed of N unmanned ships to not exceed the maximum communication radius L of the unmanned ship, set the increment of the spacing configuration to a, then the number of spacing configuration sequences is And M is an integer, that is, the number of unmanned ship formations is M;

[0084] The CFD numerical simulation technology is used to perform numerical simulation on the formation of M unmanned ships, and the total resistance of M formations is obtained, that is, the total resistance of the formation can be calculated using the existing CFD numerical simulation technology; wherein, the total resistance of each formation includes the single resistance generated by the N unmanned ships in the formation, and each total resistance of the formation corresponds to 2 (N-1) spacing configurations;

[0085] For any one of the M unmanned ship formations, the resistances of all the unmanned ships in the unmanned ship formation when they are self-propelled individually are added together to obtain the total self-propelled resistance. That is, the resistances generated by each unmanned ship when sailing alone are added together to obtain the total self-propelled resistance. The total formation resistance is divided by the total self-propelled resistance to achieve dimensionless processing, and the result obtained is the overall energy consumption of the formation in the current formation.

[0086] S13: Taking M spacing configuration sequences as input variables and the overall energy consumption of the formation corresponding to the M spacing configuration sequences as output variables, a response surface approximate model is constructed for the total formation resistance corresponding to the M unmanned ship formations. The energy consumption calculation model of the unmanned ship formation is obtained by using a high-order response surface fitting method. Its mathematical form is as follows:

[0087] En=α1d1+α2d2+,···,+α N-1 d N-1 +β1s1+β2s2+,···,+β N-1 s N-1

[0088] In the formula, α1, α2, …, α N-1 and β1,β2,…,β N-1 are the regression coefficients of the i-th longitudinal offset and the i-th lateral deviation respectively; En is the total energy consumption of the formation, referred to as the energy consumption value.

[0089] S2: constructing a formation energy consumption historical data matrix and a formation energy consumption current data matrix based on the formation total resistance and the unmanned ship formation energy consumption calculation model;

[0090] In a specific embodiment, in S2, the process of constructing a formation energy consumption historical data matrix and a formation energy consumption current data matrix based on the formation total resistance and the unmanned ship formation energy consumption calculation model is as follows:

[0091] S21: Sort the total resistances of the M formations according to the formation numbers, take the total resistances of the first M-1 formations as the historical data of formation energy consumption, take the spacing configuration sequence corresponding to the total resistances of the first M-1 formations as the unit row vector, arrange them from top to bottom in the order of the formation numbers, and take the energy consumption value corresponding to each spacing configuration sequence as the last column, so as to construct a formation energy consumption historical data matrix;

[0092] S22: The Mth formation total resistance among the M formation total resistances is used as the current data of formation energy consumption, the spacing configuration sequence corresponding to the Mth total resistance is used as the unit row vector, and the corresponding energy consumption value is used as the last column to construct the current data matrix of formation energy consumption.

[0093] S3: construct a dynamic objective function based on the unmanned ship formation energy consumption calculation model, and obtain a current dynamic data matrix based on the dynamic objective function and the current data matrix of formation energy consumption; construct a future data matrix of formation energy consumption based on the historical data matrix of formation energy consumption and the current dynamic data matrix, so as to obtain all spacing configurations and energy consumption values ​​under the whole voyage, that is, a formation energy consumption time domain database;

[0094] In a specific embodiment, in S3, a dynamic objective function is constructed based on the unmanned ship formation energy consumption calculation model, and a current dynamic data matrix is ​​obtained based on the dynamic objective function and the current data matrix of the formation energy consumption; a future data matrix of the formation energy consumption is constructed based on the historical data matrix of the formation energy consumption and the current dynamic data matrix, so as to obtain all spacing configurations and energy consumption values ​​under the whole voyage, that is, the process of the formation energy consumption time domain database is:

[0095] S31: Figure 5 As shown, an improved moving peak function is used to construct a formation dynamic environment, and the improved moving peak function is expressed as:

[0096]

[0097] Where f(x,α(t)) represents the moving peak function constructed in the form of multi-dimensional and multi-peak peak function;

[0098] H i (t), W i (t) and C i (t) represents the peak height, peak width and center position of the i-th moving peak function at time t; m is the number of peaks; x represents the spacing configuration sequence;

[0099] S32: Set the movement form of the formation dynamic environment, namely:

[0100] Assume that the peak height, peak width and center position of the moving peak function at time t+1 are in the following form:

[0101] H i(t+1)=H i (t)+height_severity i Rand(0,1)

[0102] W i (t+1)=W i (t)+width_severity i Rand(0,1)

[0103] C i (t+1)=C i (t)+ν i (t+1)

[0104] In the formula, H i (t+1), W i (t+1) and C i (t+1) represents the peak height, peak width and center position of the i-th moving peak function at time t+1; height_severity i and width_severity i Respectively represent the degree of change in peak height and peak width of the i-th peak; ν i represents a moving vector; Rand(0,1) represents a random number generated between [0,1];

[0105] The moving direction and moving distance of the center position of the moving peak function and the degree of change of the peak height and peak width are set, so as to realize the dynamic change of the moving peak function;

[0106] S33: construct dynamic objective function;

[0107] The energy consumption calculation model of the unmanned ship formation is multiplied by the improved moving peak function to construct a dynamic function of the formation energy consumption, that is, a dynamic objective function. The form of the dynamic objective function is as follows:

[0108] E(x,α(t))=f(x,α(t))·En Δ

[0109] Where E(x,α(t)) represents the dynamic objective function, f(x,α(t)) is the improved moving peak function, En is the energy consumption calculation model of the unmanned ship formation, and Δ is the time-varying parameter;

[0110] S34: Substitute the current data matrix of formation energy consumption into the dynamic objective function to obtain a current dynamic data matrix;

[0111] S35: Taking the formation energy consumption history data matrix and the spacing configuration sequence in the current dynamic data matrix as input variables, using a predictor model to predict the spacing configuration sequence at a future moment and output it. The predictor model is as follows:

[0112]

[0113] Where, X d ' s The sequence of spacing configurations for future moments predicted by the predictor model; The spacing configuration sequence reserved for the environment at the previous moment; λ is the predictor model coefficient; X ds is the spacing configuration sequence in the formation energy consumption historical data matrix and the formation energy consumption current data matrix input into the predictor model; P is the training parameter of the predictor model; Δt is the number of training environments;

[0114] Among them, the training parameter P of the predictor model is iteratively trained according to the preset environmental gradient in the current environment, and then the trained P is fed back to the predictor model to change X d ' s The result; then the current moment environment is switched to the next moment environment, and P performs the next self-training; this is repeated until the number of environment switches reaches the preset upper limit of the number of environments, the training parameters P of the predictor model stop self-training, the predictor model prediction process stops, and all X are output d ' s ;

[0115] The predicted spacing configuration sequences of all future moments are taken as unit row vectors, and arranged in order from top to bottom according to the order of environment numbers. The energy consumption value corresponding to each spacing configuration sequence is calculated according to the energy consumption calculation model of the unmanned ship formation, and the corresponding energy consumption value is taken as the last column in turn, thereby constructing the future data matrix of the formation energy consumption;

[0116] S36: Arrange the formation energy consumption historical data matrix, the current dynamic data matrix and the future data matrix in chronological order, so as to obtain a formation energy consumption time domain database.

[0117] S4: Preliminarily determine reasonable local optimal decision variables and local minimum switching costs corresponding to the reasonable local optimal decision variables from the formation energy consumption time domain database based on the time domain robust optimization algorithm and the dynamic objective function, wherein the decision variables are the spacing configuration of the unmanned ship formation;

[0118] In a specific embodiment, in S4, the process of preliminarily determining a reasonable local optimal decision variable from the formation energy consumption time domain database based on the time domain robust optimization algorithm and the dynamic objective function, and the local minimum switching cost corresponding to the reasonable local optimal decision variable is:

[0119] S41: Design a formula for calculating the cost of switching formations

[0120] The switching cost is measured by the Euclidean distance between two adjacent groups of decision variables in the formation energy consumption time domain database, thereby characterizing the degree of change of decision variables in different environments. The decision variables are the spacing configuration sequence. The calculation formula of the formation switching cost is as follows:

[0121]

[0122] Where, d i and represents the i-th longitudinal offset before and after the environment switch, s i and represents the i-th lateral deviation before and after the environment switch, || ||2 represents the second norm, N is the number of formation members, M is the number of formations, K is the total number of environments, and Ec is the formation switching cost;

[0123] S42: Figure 6 As shown, the design execution environment switching operation judgment process:

[0124] The execution environment switching operation judgment process is as follows: determine whether to execute the environment switching operation according to the dynamic objective function value corresponding to the decision variable at time t, that is, if E(x, α(t)) ≥ 0.9E(x0, α(t0)), it is determined that the decision variable at time t does not meet the current environment requirements, then execute the environment switching operation process, otherwise the execution environment switching operation is terminated, and the time domain robust optimization algorithm outputs the decision variable at the current moment, which is the initial local optimal decision variable; wherein x0 represents the decision variable under the initial environment, t0 represents the initial moment; E represents the dynamic function of formation energy consumption;

[0125] The process of the environment switching operation is as follows: updating the environment conditions at time t to the environment conditions at time t+1, updating the environment parameters to the moving peak function parameters, and adjusting the values ​​of the environment parameters according to the dynamic change form of the moving peak function; obtaining the decision variables and their energy consumption values ​​at time t+1 from the future data matrix of the formation energy consumption, and if all the future data of the formation energy consumption have been called, executing S43, otherwise calculating the moving peak function value of the decision variable at time t+1 according to the moving peak function, and then obtaining the dynamic objective function value of the decision variable at time t+1 in the new environment, and replacing the dynamic objective function value at time t with the dynamic objective function value at time t+1, and updating it to E(x, α(t)), and replacing the decision variable at time t with the decision variable at time t+1;

[0126] S43: The environment switching operation is terminated, and the time-domain robust optimization algorithm outputs the decision variables at the last moment, which are the initial local optimal decision variables;

[0127] S44: Check the rationality of the initial local optimal decision variables

[0128] The initial local optimal decision variable is tested according to the number of environment switching, and the testing rule is: if the number of environment switching corresponding to the initial local optimal decision variable is 1 or the set maximum number of environments, and the set maximum number of environments is not 1, then the initial local optimal decision variable is determined to be unreasonable. At this time, the unreasonable initial local optimal decision variable is eliminated and S45 is executed; otherwise, a reasonable local optimal decision variable is obtained;

[0129] Specifically, when the number of environment switches corresponding to the local optimal decision variable is equal to 1, the characterization environment switches from time to time (the formation needs to be switched in each environment). Therefore, in practical applications, the formation is not timely, so the local optimal decision variable is judged to be unreasonable. When the number of environment switches corresponding to the local optimal decision variable is equal to the maximum number of environments, the characterization environment does not switch, indicating that the formation navigation environment is not dynamic in practical applications. Therefore, the local optimal decision variable is judged to be unreasonable. After eliminating unreasonable variables, the final local optimal decision variable is used for subsequent population evolution.

[0130] S45: changing the spacing configuration sequence and the corresponding energy consumption value in the current dynamic data matrix to the spacing configuration sequence and the corresponding energy consumption value at the previous moment, that is, the spacing configuration sequence and the corresponding energy consumption value in the formation energy consumption history data matrix, and then repeatedly executing S41-S44 until a reasonable local optimal decision variable is obtained or the formation energy consumption history data is called up, then terminating the calculation;

[0131] S46: According to the formation switching cost formula, the switching cost between the reasonable local optimal decision variable and the decision variable at time t is calculated, which is the local minimum switching cost.

[0132] S5: The wild dog optimization algorithm is used to optimize the reasonable local optimal decision variables, and finally the global optimal decision variables are obtained, that is, the optimal spacing configuration sequence that takes into account both minimum energy consumption and minimum switching cost; the unmanned ship formation constructed based on the optimal spacing configuration sequence is the unmanned ship energy-saving formation.

[0133] In a specific embodiment, in S5, the wild dog optimization algorithm is used to optimize the reasonable local optimal decision variables, and finally the global optimal decision variables are obtained, that is, the process of the optimal spacing configuration sequence that takes into account the minimum energy consumption and the minimum switching cost is:

[0134] S51: Calculate the fitness value;

[0135] The wild dog optimization algorithm is used to randomly generate several decision variables of the same dimension according to the sequence dimension of the reasonable local optimal decision variable, which is called individuals. All individuals are arranged from top to bottom in the order of generation to form a population matrix, which is called the initial population. Among them, the reasonable local optimal decision variable is placed in the first row of the initial population as the current optimal individual. The energy consumption value of the current optimal individual is calculated according to the formation energy consumption calculation model. The energy consumption value of the current optimal individual is added to the local minimum switching cost and the inverse is taken to obtain the current fitness value. The calculation formula is as follows:

[0136]

[0137] Where ω1 and ω2 are weight coefficients obtained after normalization, and their sum is 1; En is the total energy consumption of the formation, that is, the energy consumption value; Ec is the formation switching cost;

[0138] S52: The process of evolving the reasonable local optimal decision variables using the wild dog optimization algorithm is as follows:

[0139] S521: The population evolution operation in the wild dog optimization algorithm includes four methods: group attack, persecution behavior, scavenging behavior and wild dog survival rate. The evolution principle is set to the principle of minimizing the total energy consumption of the formation and the formation switching cost at the same time, that is, the principle of maximizing the fitness value;

[0140] S522: Execute the first round of evolution based on the wild dog optimization algorithm, perform population evolution operation on the first row of individuals in the initial population, obtain new individuals after evolution, that is, the spacing configuration sequence of the unmanned ship formation, calculate the formation switching cost of the new individuals and the original individuals through the formation switching cost calculation formula, calculate the energy consumption value of the new individuals through the formation energy consumption calculation model, and then obtain the fitness value of the new individuals;

[0141] S523: Determine whether the fitness value of the new individual is greater than the fitness value of the original individual before evolution. If so, determine that the new individual is superior, and replace the current optimal individual with the new individual, and replace the current fitness value with the fitness value of the new individual; otherwise, determine that the original individual is superior, and retain the current optimal individual and the current fitness value;

[0142] S524: According to the step content of S522, the population evolution operation is performed on the remaining rows of individuals in the initial population in turn, and then S523 is repeated to perform the judgment operation until all individuals in the initial population are evolved. Finally, the current optimal individual obtained by judgment is the global optimal decision variable, and the obtained global optimal decision variable is the optimal spacing configuration sequence that takes into account both minimum energy consumption and minimum switching cost.

[0143] In this embodiment, a simulation experiment is carried out to verify the effectiveness of the method provided in this embodiment. A four-formation sailing formation is constructed for the unmanned ship cluster. The four ships are, in order from front to back, the pilot ship, follower ship No. 1, follower ship No. 2 and follower ship No. 3; the spacing is configured into three groups, namely, the lateral deviation s1 and the longitudinal offset d1 between the pilot ship and the follower ship No. 1, the lateral deviation s2 and the longitudinal offset d2 between the follower ship No. 1 and the follower ship No. 2, and the lateral deviation s3 and the longitudinal offset d3 between the follower ship No. 2 and the follower ship No. 3; the single resistances of the four unmanned ships in the formation are R1, R2, R3 and R4 respectively, and the total formation resistance of the current formation is R T =R1+R2+R3+R4. Since the four ships have the same configuration, the self-propulsion resistance of each of the four ships is R. The overall energy consumption of the formation in the current formation is R T / (4R). The above six spacing configurations are used as decision variables, and the spacing configuration values ​​in the initial formation are the initial values ​​of the decision variables, which are D1∈[2,5], D2∈[2.1,3.1], D3∈[2.2,3.3], S1∈[0.5,1.7], S2∈[0.3,1.3], S3∈[0.2,0.9], the maximum number of environments is 13, and the environment gradient is 0.7.

[0144] Specifically, Figure 2-Figure 4 As shown in , after the formation dynamic adjustment method is adopted, the formation obtained is the energy-saving formation after the formation adjustment. Compare the energy-saving formation after the formation adjustment with the energy-saving formation before the formation adjustment, the change in energy consumption is the energy-saving rate, and the change in switching cost is the scheduling distance. Figure 7 As shown, the energy consumption of the energy-saving formation after dynamic formation adjustment is reduced by 10% compared with the energy consumption of the initial formation, and the energy saving rate of the energy-saving formation after the formation adjustment is determined to be 10%. This energy-saving formation can be considered to be the most energy-saving formation in the entire navigation cycle, and the spacing configuration corresponding to this energy-saving formation is the spatial position distribution of the most energy-saving formation.

[0145] Finally, it should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit it. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that they can still modify the technical solutions described in the aforementioned embodiments, or replace some or all of the technical features therein with equivalents. However, these modifications or replacements do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. A method for adjusting the dynamic formation of an unmanned ship fleet considering switching costs, characterized in that: The specific steps include: S1: Obtaining a number of total formation resistances, and constructing an unmanned ship formation energy consumption calculation model based on the total formation resistances, wherein the unmanned ship formation energy consumption calculation model is used to calculate the total formation energy consumption; S2: constructing a formation energy consumption historical data matrix and a formation energy consumption current data matrix based on the formation total resistance and the unmanned ship formation energy consumption calculation model; S3: construct a dynamic objective function based on the unmanned ship formation energy consumption calculation model, and obtain a current dynamic data matrix based on the dynamic objective function and the current data matrix of formation energy consumption; construct a future data matrix of formation energy consumption based on the historical data matrix of formation energy consumption and the current dynamic data matrix, so as to obtain all spacing configurations and energy consumption values ​​under the whole voyage, that is, a formation energy consumption time domain database; In S3, a dynamic objective function is constructed based on the unmanned ship formation energy consumption calculation model, and a current dynamic data matrix is ​​obtained based on the dynamic objective function and the current data matrix of the formation energy consumption; a future data matrix of the formation energy consumption is constructed based on the historical data matrix of the formation energy consumption and the current dynamic data matrix, so as to obtain all spacing configurations and energy consumption values ​​under the whole voyage, that is, the process of the formation energy consumption time domain database is as follows: S31: constructing a formation dynamic environment using an improved moving peak function, wherein the improved moving peak function is expressed as: Where f(x,α(t)) represents the moving peak function constructed in the form of multi-dimensional and multi-peak peak function; H i (t), W i (t) and C i (t) represents the peak height, peak width and center position of the i-th moving peak function at time t; m is the number of peaks; x represents the spacing configuration sequence; S32: Set the movement form of the formation dynamic environment, namely: Assume that the peak height, peak width and center position of the moving peak function at time t+1 are in the following form: H i (t+1)=H i (t)+height_severity i ·Rand(0,1) W i (t+1)=W i (t)+width_severity i ·Rand(0,1) C i (t+1)=C i (t)+ν i (t+1) In the formula, H i (t+1), W i (t+1) and C i (t+1) represents the peak height, peak width and center position of the i-th moving peak function at time t+1; height_severity i and width_severity i Respectively represent the degree of change in peak height and peak width of the i-th peak; ν i represents a moving vector; Rand(0,1) represents a random number generated between [0,1]; The moving direction and moving distance of the center position of the moving peak function and the degree of change of the peak height and peak width are set, so as to realize the dynamic change of the moving peak function; S33: construct dynamic objective function; The energy consumption calculation model of the unmanned ship formation is multiplied by the improved moving peak function to construct a dynamic function of the formation energy consumption, that is, a dynamic objective function. The form of the dynamic objective function is as follows: E(x,α(t))=f(x,α(t))·En Δ Where E(x,α(t)) represents the dynamic objective function, f(x,α(t)) is the improved moving peak function, En is the energy consumption calculation model of the unmanned ship formation, and Δ is the time-varying parameter; S34: Substitute the current data matrix of formation energy consumption into the dynamic objective function to obtain a current dynamic data matrix; S35: Taking the formation energy consumption history data matrix and the spacing configuration sequence in the current dynamic data matrix as input variables, using a predictor model to predict the spacing configuration sequence at a future moment and output it. The predictor model is as follows: In the formula, X′ ds The sequence of spacing configurations for future moments predicted by the predictor model; The spacing configuration sequence reserved for the environment at the previous moment; λ is the predictor model coefficient; X ds is the spacing configuration sequence in the formation energy consumption historical data matrix and the formation energy consumption current data matrix input into the predictor model; P is the training parameter of the predictor model; Δt is the number of training environments; Among them, the training parameter P of the predictor model is iteratively trained according to the preset environmental gradient in the current environment, and then the trained P is fed back to the predictor model to change X′ ds The result; then the current environment is switched to the next environment, and P performs the next self-training; this is repeated until the number of environment switches reaches the preset upper limit of the number of environments, the training parameters P of the predictor model stop self-training, the predictor model prediction process stops, and all X′ are output ds ; The predicted spacing configuration sequences of all future moments are taken as unit row vectors, and arranged in order from top to bottom according to the order of environment numbers. The energy consumption value corresponding to each spacing configuration sequence is calculated according to the energy consumption calculation model of the unmanned ship formation, and the corresponding energy consumption value is taken as the last column in turn, thereby constructing the future data matrix of the formation energy consumption; S36: arranging the formation energy consumption historical data matrix, the current dynamic data matrix and the future data matrix in chronological order, thereby obtaining a formation energy consumption time domain database; S4: Preliminarily determine reasonable local optimal decision variables and local minimum switching costs corresponding to the reasonable local optimal decision variables from the formation energy consumption time domain database based on the time domain robust optimization algorithm and the dynamic objective function, wherein the decision variables are the spacing configuration sequence of the unmanned ship formation; S5: The wild dog optimization algorithm is used to optimize the reasonable local optimal decision variables, and finally the global optimal decision variables are obtained, that is, the optimal spacing configuration sequence that takes into account both minimum energy consumption and minimum switching cost; the unmanned ship formation constructed based on the optimal spacing configuration sequence is the unmanned ship energy-saving formation.

2. The method for dynamic formation adjustment of unmanned ship formation considering switching cost according to claim 1 is characterized in that: In S1, the process of obtaining a plurality of formation total resistances and constructing an unmanned ship formation energy consumption calculation model based on the formation total resistances is as follows: S11: Assume an unmanned ship formation consisting of N unmanned ships. Each formation is determined by N-1 sets of spacing configurations. The corresponding number of spacing configurations is 2(N-1). Each formation corresponds to a spacing configuration sequence, recorded as (d1, d2, d3, ···d i ,···,d N-1 ,s1,s2,s3,···s i ,···,s N-1 ), where d i is the i-th longitudinal offset before the environment switch, s i is the i-th lateral deviation before the environment switch; S12: Set the lateral deviation and longitudinal offset of each unmanned ship in the unmanned ship formation composed of N unmanned ships to not exceed the maximum communication radius L of the unmanned ship, set the increment of the spacing configuration to a, then the number of spacing configuration sequences is And M is an integer, that is, the number of unmanned ship formations is M; The CFD numerical simulation technology is used to simulate the formation of M unmanned ships and the total resistance of the M formations is obtained. For any unmanned ship formation among the M unmanned ship formations, the resistance of all unmanned ships in the unmanned ship formation when a single ship is self-propelled is added to obtain the total self-propelled resistance, and the total formation resistance is divided by the total self-propelled resistance to achieve dimensionless processing, and the result obtained is the overall energy consumption of the formation in the current formation; S13: Taking M spacing configuration sequences as input variables and the overall energy consumption of the formation corresponding to the M spacing configuration sequences as output variables, a response surface approximate model is constructed for the total formation resistance corresponding to the M unmanned ship formations. The energy consumption calculation model of the unmanned ship formation is obtained by using a high-order response surface fitting method. Its mathematical form is as follows: En=α1d1+α2d2+,···,+α N-1 d N-1 +β1s1+β2s2+,···,+β N-1 s N-1 In the formula, α1, α2, …, α N-1 and β1,β2,…,β N-1 are the regression coefficients of the i-th longitudinal offset and the i-th lateral deviation respectively; En is the total energy consumption of the formation, referred to as the energy consumption value.

3. The method for dynamic formation adjustment of unmanned ship formation considering switching cost according to claim 2 is characterized in that: In S2, the process of constructing the formation energy consumption historical data matrix and the formation energy consumption current data matrix based on the formation total resistance and the unmanned ship formation energy consumption calculation model is as follows: S21: Sort the total resistances of the M formations according to the formation numbers, take the total resistances of the first M-1 formations as the historical data of formation energy consumption, take the spacing configuration sequence corresponding to the total resistances of the first M-1 formations as the unit row vector, arrange them from top to bottom in the order of the formation numbers, and take the energy consumption value corresponding to each spacing configuration sequence as the last column, so as to construct a formation energy consumption historical data matrix; S22: The Mth formation total resistance among the M formation total resistances is used as the current data of formation energy consumption, the spacing configuration sequence corresponding to the Mth total resistance is used as the unit row vector, and the corresponding energy consumption value is used as the last column to construct the current data matrix of formation energy consumption.

4. The method for dynamic formation adjustment of unmanned ship formation considering switching cost according to claim 3 is characterized in that: In S4, based on the time domain robust optimization algorithm and the dynamic objective function, a process of preliminarily determining a reasonable local optimal decision variable from the formation energy consumption time domain database and a local minimum switching cost corresponding to the reasonable local optimal decision variable is as follows: S41: Design a formula for calculating the cost of switching formations The switching cost is measured by the Euclidean distance between two adjacent groups of decision variables in the formation energy consumption time domain database, thereby characterizing the degree of change of decision variables in different environments. The decision variables are the spacing configuration sequence. The calculation formula of the formation switching cost is as follows: Where, d i and represents the i-th longitudinal offset before and after the environment switch, s i and represents the i-th lateral deviation before and after the environment switch, ‖‖2 represents the second norm, N is the number of formation members, M is the number of formations, K is the total number of environments, and Ec is the formation switching cost; S42: Design execution environment switching operation judgment process: The execution environment switching operation judgment process is as follows: determine whether to execute the environment switching operation according to the dynamic objective function value corresponding to the decision variable at time t, that is, if E(x, α(t)) ≥ 0.9E(x0, α(t0)), it is determined that the decision variable at time t does not meet the current environment requirements, then execute the environment switching operation process, otherwise the execution environment switching operation is terminated, and the time domain robust optimization algorithm outputs the decision variable at the current moment, which is the initial local optimal decision variable; wherein x0 represents the decision variable under the initial environment, t0 represents the initial moment; E represents the dynamic function of formation energy consumption; The process of the environment switching operation is as follows: updating the environment conditions at time t to the environment conditions at time t+1, updating the environment parameters to the moving peak function parameters, and adjusting the values ​​of the environment parameters according to the dynamic change form of the moving peak function; obtaining the decision variables and their energy consumption values ​​at time t+1 from the future data matrix of the formation energy consumption, and if all the future data of the formation energy consumption have been called, executing S43, otherwise calculating the moving peak function value of the decision variable at time t+1 according to the moving peak function, and then obtaining the dynamic objective function value of the decision variable at time t+1 in the new environment, and replacing the dynamic objective function value at time t with the dynamic objective function value at time t+1, and updating it to E(x, α(t)), and replacing the decision variable at time t with the decision variable at time t+1; S43: The environment switching operation is terminated, and the time-domain robust optimization algorithm outputs the decision variables at the last moment, and uses the decision variables at the last moment as the initial local optimal decision variables; S44: Check the rationality of the initial local optimal decision variables The initial local optimal decision variable is tested according to the number of environment switching, and the testing rule is: if the number of environment switching corresponding to the initial local optimal decision variable is 1 or the set maximum number of environments, and the set maximum number of environments is not 1, then the initial local optimal decision variable is determined to be unreasonable. At this time, the unreasonable initial local optimal decision variable is eliminated and S45 is executed; otherwise, a reasonable local optimal decision variable is obtained; S45: changing the spacing configuration sequence and the corresponding energy consumption value in the current dynamic data matrix to the spacing configuration sequence and the corresponding energy consumption value at the previous moment, that is, the spacing configuration sequence and the corresponding energy consumption value in the formation energy consumption history data matrix, and then repeatedly executing S41-S44 until a reasonable local optimal decision variable is obtained or the formation energy consumption history data is called up, then terminating the calculation; S46: According to the formation switching cost formula, the switching cost between the reasonable local optimal decision variable and the decision variable at time t is calculated, which is the local minimum switching cost.

5. The method for dynamic formation adjustment of unmanned ship formation considering switching cost according to claim 4 is characterized in that: In S5, the wild dog optimization algorithm is used to optimize the reasonable local optimal decision variables, and finally the global optimal decision variables are obtained, that is, the process of the optimal spacing configuration sequence that takes into account the minimum energy consumption and the minimum switching cost is: S51: Calculate the fitness value; The wild dog optimization algorithm is used to randomly generate several decision variables of the same dimension according to the sequence dimension of the reasonable local optimal decision variable, which is called individuals. All individuals are arranged from top to bottom in the order of generation to form a population matrix, which is called the initial population. Among them, the reasonable local optimal decision variable is placed in the first row of the initial population as the current optimal individual. The energy consumption value of the current optimal individual is calculated according to the formation energy consumption calculation model. The energy consumption value of the current optimal individual is added to the local minimum switching cost and the inverse is taken to obtain the current fitness value. The calculation formula is as follows: Where ω1 and ω2 are weight coefficients obtained after normalization, and their sum is 1; En is the total energy consumption of the formation, that is, the energy consumption value; Ec is the formation switching cost; S52: The process of evolving the reasonable local optimal decision variables using the wild dog optimization algorithm is as follows: S521: Set the evolution principle as the principle of minimizing both the total energy consumption of the formation and the cost of switching the formation, that is, the principle of maximizing the fitness value; S522: Execute the first round of evolution based on the wild dog optimization algorithm, perform population evolution operation on the first row of individuals in the initial population, obtain new individuals after evolution, that is, the spacing configuration sequence of the unmanned ship formation, calculate the formation switching cost of the new individuals and the original individuals through the formation switching cost calculation formula, calculate the energy consumption value of the new individuals through the formation energy consumption calculation model, and then obtain the fitness value of the new individuals; S523: Determine whether the fitness value of the new individual is greater than the fitness value of the original individual before evolution. If so, determine that the new individual is superior, and replace the current optimal individual with the new individual, and replace the current fitness value with the fitness value of the new individual; otherwise, determine that the original individual is superior, and retain the current optimal individual and the current fitness value; S524: According to the step content of S522, the population evolution operation is performed on the remaining rows of individuals in the initial population in turn, and then S523 is repeated to perform the judgment operation until all individuals in the initial population are evolved. Finally, the current optimal individual obtained by judgment is the global optimal decision variable, and the obtained global optimal decision variable is the optimal spacing configuration sequence that takes into account both minimum energy consumption and minimum switching cost.

Images