Low-cost flower sea plant screening method and landscape scheme generation method and system

Abstract

The invention discloses a low-cost flower sea plant screening method and a landscape scheme generation method. The low-cost flower sea plant screening method comprises the following steps: acquiring indexes for evaluating low-cost flower sea plants; determining the weight of each index; training to obtain a low-cost flower sea plant screening model; and forming a variety library of the flower sea plants in the region according to the variety information of the flower sea plants in the to-be-screened region. Constructing a flower sea plant landscape effect evaluation system, collecting a flower sea plant landscape plan, scoring the plan effect according to the evaluation system, and constructing a flower sea landscape configuration training set; training the model to obtain a landscape scheme generation model; outputting coordinates by adopting a particle swarm algorithm, and generating a planar graph through plant picture information. According to the method, an evaluation system is established according to a low-cost flower sea plant screening target, a low-cost flower sea plant screening model is constructed through a neural network, a low-cost flower sea plant recommendation list is obtained, a low-cost flower sea landscape configuration scheme is further generated according to site basic conditions and environmental conditions, and a decision basis is provided for flower sea landscape construction.

Description

Technical Field

[0001] The present invention belongs to the field of plant landscape, and in particular to the technical field of low-cost flower sea plant screening and landscape configuration scheme generation. Background Art

[0002] Unrestrained human resource demands and urbanization have led to serious deterioration of the urban environment and frequent ecological problems. In response to this, my country has proposed to build a "resource-saving and environmentally friendly" city. Against this background, the Ministry of Housing and Urban-Rural Development has called for the construction of a "saving" landscape. As an important type of leisure and sightseeing project during the booming leisure tourism in recent years, the low-maintenance, low-cost and sustainable landscape construction of flower sea landscape has gradually become an important content of research. Low-cost flower sea refers to a construction concept that reduces the long-term cost of the landscape by reducing the maintenance and replacement of materials, saving energy consumption and creating economic compensation during the maintenance stage of landscape gardening. The research on low-cost flower sea landscape configuration technology has strong practical significance and practical value under the background of the times.

[0003] The construction of low-cost flower sea landscape involves many elements and complex relationships. Especially when the amount of data to be processed is huge and the overall complex system has complex relationships, it is no longer possible to construct a low-cost flower sea landscape system through traditional methods. In response to these problems, some scholars have conducted in-depth research, such as: using Visual C++6.0 and SOL Server database hybrid programming technology combined with fuzzy comprehensive evaluation method to develop a low-cost garden plant selection system; using the hierarchical analysis method (AHP) to construct a comprehensive screening system for maintenance-free flower border plants and a low-cost garden plant screening system.

[0004] Although the fuzzy comprehensive evaluation method and the hierarchical analysis method have solved the above problems to a certain extent, these two methods mainly rely on the experience of experts to determine the weight values ​​of various indicators, which are highly subjective and have low interpretability, resulting in the screening results being relatively unscientific and objective. Therefore, it is necessary to further study the indicator factors and weights of various indicators in the low-cost plant evaluation system to reduce the interference of human subjective factors and make the evaluation results more objective and scientific. Summary of the invention

[0005] Purpose of the invention: The purpose of the present invention is to provide a low-cost flower sea plant screening method and system to address the deficiencies of the prior art, to construct a low-cost flower sea evaluation system, and to use computer deep learning to screen out low-cost flower sea plant species;

[0006] Another object of the present invention is to provide a method and system for generating a landscape plan, using the selected low-cost flower sea plant species to form a low-cost flower sea landscape configuration plan.

[0007] In order to achieve the above-mentioned purpose, the present invention provides a low-cost method for screening sea of ​​flowers plants, comprising the following steps:

[0008] S1. Obtain indicators for evaluating low-cost flower sea plants;

[0009] S2. Determine the weight of each indicator and construct a comprehensive evaluation data set;

[0010] S3. Based on the weights of each indicator and the comprehensive evaluation data set, a BP neural network framework is used to train and obtain a low-cost flower sea plant screening model;

[0011] S4. Use the information of flower plant varieties in the area to be screened as the input of the trained low-cost flower plant screening model, output low-cost flower plants that meet the requirements, and form a flower plant variety library in the area.

[0012] A further preferred technical solution of the present invention is that the step S1 of obtaining the index for evaluating low-cost flower sea plants is specifically performed as follows:

[0013] Based on the literature, the key factors for evaluating low-cost flower plants were determined. Then, through public and expert surveys, the main indicators of concern under different criteria were determined to form an indicator system for evaluating low-cost plants, including:

[0014] (1) The main indicators of concern under the low maintenance criteria are maintenance needs, replacement frequency and economy;

[0015] (2) The main indicators of interest under the ornamental criteria are the length of the ornamental period, flower color, plant shape, and leaf shape and color;

[0016] (3) The main indicators of concern under the adaptability criterion are tolerance to barrenness, shade tolerance, drought tolerance, waterlogging tolerance, disease and pest resistance, reproductive capacity and locality.

[0017] Preferably, the step S2 of determining the weight of each indicator and constructing a comprehensive evaluation data set includes:

[0018] S21. Determine the weight of each indicator using the entropy weight method;

[0019] S22. Construct a comprehensive evaluation dataset.

[0020] Preferably, the entropy weight method is used to determine the weight of each indicator in step S21, and the specific method is:

[0021] S211, let e j is the entropy value of the jth evaluation index, then the entropy value e j The calculation process is:

[0022]

[0023] In the formula, f ij is the characteristic weight of the i-th system under the j-th index; x ij is the observed data of the jth indicator in the i-th system, i=1,2,…,n; j=1,2,…,m; is the sum of all system observation data of the jth indicator;

[0024] S212, let w j * is the entropy weight of the jth evaluation index, then the entropy weight of the index is as follows:

[0025]

[0026] In the formula, e i is the entropy value of the i-th indicator;

[0027] S213, let P i is the evaluation score of the ith plant. According to the linear weighted comprehensive evaluation formula, the comprehensive evaluation score is:

[0028]

[0029] In the formula, p ij is the normalized score of the evaluation index; W j * is the weight of the jth evaluation index.

[0030] Preferably, the comprehensive evaluation data set is constructed in step S22 by:

[0031] Through a questionnaire survey, landscape design practitioners were asked to score each indicator from 1 to 10 points. Qualified questionnaires were retained as the original sample data, and the training set and test set for training the BP neural network model were constructed in a ratio of 9:1.

[0032] Preferably, in step S3, according to the weights of each indicator and the comprehensive evaluation data set, a BP neural network framework is used to train and obtain a low-cost flower sea plant screening model, and the specific method is:

[0033] S31. Use the low-cost plant evaluation index as the input layer of the BP neural network, and the low-cost flower plant evaluation score as the output layer of the BP neural network. Determine the number of implicit nodes of the BP neural network according to the trial and error method and formula, so as to establish the topological structure of the BP neural network.

[0034] The number of input, hidden and output nodes of the BP neural network are N, L and M respectively, and the input vector is X = [x 0 ,x 1 ,…,x N-1], the weights between hidden layer point j and input layer point i and output layer node k are V ij and W jk , the thresholds of the output layer and the hidden layer are θ k and φ j , the output calculation formula of the hidden layer nodes and the output layer is:

[0035]

[0036] Calculate y k With the target output d k Deviation:

[0037] δ k =(d k -y k )y k 1-y k );

[0038]

[0039] The weights are resized to:

[0040]

[0041]

[0042] The weight adjustment formula is:

[0043] W jk (n+1)=W jk (n)+ΔW jk (n)+μΔW jk (n-1);

[0044] V ij (n+1)=V ij (n)+ΔV ij (n)+μΔV ij (n-1);

[0045] S32. The evaluation levels are divided and the comprehensive evaluation scores are judged. The comprehensive evaluation scores ≥70% are classified as Class I, and the comprehensive evaluation scores <70% are classified as Class II. Class I plants are defined as low-cost plants;

[0046] S33. Use the comprehensive evaluation data set for training, construct a mapping from indicators to evaluation levels, directly output whether it is a low-cost flower plant, and obtain a low-cost flower plant screening model.

[0047] Another aspect of the present invention provides a low-cost flower sea plant screening system, comprising:

[0048] A data acquisition module, used to obtain indicators for evaluating low-cost flower sea plants;

[0049] The first data set construction module is used to determine the weight of each indicator and construct a comprehensive evaluation data set;

[0050] The first model training module is used to train a low-cost flower sea plant screening model using a BP neural network framework based on the weights of each indicator and a comprehensive evaluation data set;

[0051] The evaluation module is used to use the information of flower plant varieties in the area to be screened as the input of the trained low-cost flower plant screening model, output low-cost flower plants that meet the requirements, and form a flower plant variety library in the area.

[0052] Another aspect of the present invention provides a method for generating a landscape scheme, comprising the following steps:

[0053] S5. Based on the flower sea plant variety library of the area to be generated for the landscape plan obtained by the above-mentioned low-cost flower sea plant screening method, take photos of individual plants to form a low-cost flower sea plant library in the area;

[0054] S6. Construct a flower sea plant landscape effect evaluation system, collect flower sea plant landscape CAD plan drawings, use expert scoring method, score the collected plan drawing effects according to the evaluation system, obtain the average score as the final score, and construct a flower sea landscape configuration training set;

[0055] S7, using the training set, inputting the plan and site information into the convolutional neural network, training the model, and obtaining a landscape plan generation model;

[0056] S8. Use particle swarm algorithm to output coordinates and generate a plan view through the coordinate points and plant picture information in the library.

[0057] Preferably, in step S8, the particle swarm algorithm is used to output coordinates, and a plan view is generated through the coordinate points and the plant picture information in the library. The specific method is:

[0058] S81. Collect basic information about the plant configuration site, including the size, shape and design objectives of the site; collect environmental information, including the climate and soil conditions of the site;

[0059] S82, taking the collected basic information and environmental information of the plant configuration site and the low-cost flower sea plant library in the area as input, setting the dimension of the particle swarm algorithm to the number of variety libraries*3, and outputting the coordinates (x, y) of the center point of the picture and the number of selected plants;

[0060] When the number is 0, the corresponding coordinate is (-1, -1), indicating that the plant is not selected; when the number is non-zero, the coordinate is the position of the plant, and the set of coordinate points of the plant image is obtained;

[0061] S83. Generate a plan using coordinate points and plant image information, input the plan and site information into a landscape plan generation model, set the number of iterations, and repeat steps S81 to S82 if the number of iterations is lower than the number of iterations. Otherwise, output the plan with the best score as the final result.

[0062] In another aspect, the present invention provides a landscape scheme generation system, comprising:

[0063] A library generation module is used to take photos of individual plants according to the flower sea plant variety library of the landscape plan area to be generated obtained by the above-mentioned low-cost flower sea plant screening method to form a low-cost flower sea plant library in the area;

[0064] The second data set construction module is used to construct a flower sea plant landscape effect evaluation system and collect flower sea plant landscape CAD plan drawings. Expert scoring is used to score the collected plan drawing effects according to the evaluation system, and the average score is obtained as the final score to construct a flower sea landscape configuration training set.

[0065] The second model training module is used to use the training set to input the plan and site information into the convolutional neural network, train the model, and obtain a landscape plan generation model;

[0066] The landscape plan generation module is used to output coordinates using a particle swarm algorithm and generate a plan view through coordinate points and plant image information in the gallery.

[0067] Beneficial effects: The present invention establishes an evaluation system based on the low-cost flower sea plant screening target, constructs a low-cost flower sea plant screening model through a neural network, obtains a recommended list of low-cost flower sea plants, and further generates a low-cost flower sea landscape configuration plan based on the basic conditions of the site and the environmental conditions, providing a decision-making basis for the construction of the flower sea landscape.

[0068] Taking into account the key condition of low cost, a low-cost flower sea evaluation system is constructed, and computer deep learning is used to screen out low-cost flower sea plant species and form a low-cost flower sea landscape configuration plan, so as to realize the scientific determination of weights and the automation of model generation, so as to help planners scientifically select low-cost flower sea varieties and provide planning references for flower sea landscapes. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] Figure 1 A flow chart of the low-cost flower sea plant screening method of the present invention;

[0070] Figure 2 The present invention is a flow chart of the landscape scheme generating method. DETAILED DESCRIPTION

[0071] In order to make the purpose, technical solution and advantages of the present invention clearer, the technical solution in the present invention is clearly and completely described. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments, and they should not be understood as limitations on the present invention. 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. In the description of the present invention, it should be understood that the terms used are only for descriptive purposes and cannot be understood as indicating or implying relative importance.

[0072] Combine the following Figure 1-Figure 2 The invention describes a low-cost flower sea plant screening method, a landscape scheme generation method and a system.

[0073] Example 1: This example provides a low-cost method for screening flower sea plants. Figure 1 As shown, the following steps are included:

[0074] S1. Obtain indicators for evaluating low-cost flower sea plants;

[0075] S2. Determine the weight of each indicator and construct a comprehensive evaluation data set;

[0076] S3. Based on the weights of each indicator and the comprehensive evaluation data set, a BP neural network framework is used to train and obtain a low-cost flower sea plant screening model;

[0077] S4. Use the information of flower plant varieties in the area to be screened as the input of the trained low-cost flower plant screening model, output low-cost flower plants that meet the requirements, and form a flower plant variety library in the area.

[0078] The indicators for evaluating low-cost flower sea plants described in step S1 are obtained by screening out the key elements of low-cost flower sea plants based on literature data, and after investigating the degree of attention paid to them by the public and experts, the first-level indicator layer is evaluated by comprehensively considering the needs of flower sea landscape construction in terms of operating costs, aesthetic value, and ecological environmental adaptability, as shown in the low-cost flower sea plant screening evaluation system table in Table 1. Finally, low maintenance, landscape, and adaptability are established as the criterion layer of the model.

[0079] Table 1 Low-cost flower sea plant screening and evaluation system

[0080]

[0081]

[0082]

[0083] Step S2 determines the weight of each indicator and constructs a comprehensive evaluation data set, including:

[0084] S21. Determine the weight of each indicator using the entropy weight method;

[0085] S22. Construct a comprehensive evaluation dataset.

[0086] 4. The low-cost flower sea plant screening method according to claim 3 is characterized in that the entropy weight method is used to determine the weight of each indicator in step S21, and the specific method is:

[0087] S211, let e j is the entropy value of the jth evaluation index, then the entropy value e j The calculation process is:

[0088]

[0089] In the formula, f ij is the characteristic weight of the i-th system under the j-th index; x ij is the observed data of the jth indicator in the i-th system, i=1,2,…,n; j=1,2,…,m; is the sum of all system observation data of the jth indicator;

[0090] S212, let w j * is the entropy weight of the jth evaluation index, then the entropy weight of the index is as follows:

[0091]

[0092] In the formula, e i is the entropy value of the i-th indicator;

[0093] S213, let P i is the evaluation score of the ith plant. According to the linear weighted comprehensive evaluation formula, the comprehensive evaluation score is:

[0094]

[0095] In the formula, p ij is the normalized score of the evaluation index; W j * is the weight of the kth evaluation index.

[0096] As an objective weighting method, the entropy weight method can avoid the interference of human factors in the weights of each evaluation index and make the evaluation results more realistic. Therefore, it is necessary to establish an evaluation system, use the entropy weight method to adjust the weights, build a model through the BP neural network, screen low-cost and high-landscape value flower sea plant varieties, and automatically form a low-cost flower sea plant landscape configuration plan according to site conditions.

[0097] The specific method of constructing a comprehensive evaluation data set in step S22 is as follows:

[0098] Through questionnaire survey, landscape design practitioners were asked to rate the index system from 1 to 10 points, and 100 qualified questionnaires were retained as the original sample data. The sample data numbered 1-90 were used to train the BP neural network model, and the output result was the result of the weighted calculation of the index weight vector determined by the entropy weight method as the output value, and the sample data numbered 90-100 were used for simulation testing.

[0099] In step S3, according to the weights of each indicator and the comprehensive evaluation data set, a BP neural network framework is used to train and obtain a low-cost flower sea plant screening model. The specific method is:

[0100] S31. The low-cost plant evaluation index is used as the input layer of the BP neural network, and the low-cost flower sea plant evaluation score is used as the output layer of the BP neural network. The number of hidden nodes of the BP neural network is determined according to the trial and error method and the formula, thereby establishing the topological structure of the BP neural network:

[0101] The number of input, hidden and output nodes of the BP neural network are N, L and M respectively, and the input vector is X = [x 0 ,x 1 ,…,x N-1 ], the weights between hidden layer point j and input layer point i and output layer node k are V ij and W jk , the thresholds of the output layer and the hidden layer are θ k and φ j , the output calculation formula of the hidden layer nodes and the output layer is:

[0102]

[0103] Calculate y k With the target output d k Deviation:

[0104] δ k =(d k -y k )y k (1-y k );

[0105]

[0106] The weights are resized to:

[0107] ΔW jk (n) = ηδ k h j ;

[0108]

[0109] The weight adjustment formula is:

[0110] W jk (n+1)=W jk (n)+ΔW jk (n)+μΔW jk (n-1);

[0111] V ij (n+1)=V ij (n)+ΔV ij (n)+μΔV ij (n-1);

[0112] S32. The evaluation levels are divided and the comprehensive evaluation scores are judged. The comprehensive evaluation scores ≥70% are classified as Class I, and the comprehensive evaluation scores <70% are classified as Class II. Class I plants are defined as low-cost plants;

[0113] S33. Use the comprehensive evaluation data set for training, construct a mapping from indicators to evaluation levels, directly output whether it is a low-cost flower plant, and obtain a low-cost flower plant screening model.

[0114] As a network information processing system with high nonlinear dynamic processing capabilities, BP neural network can easily realize highly nonlinear mapping relationships between systems and is used in many fields such as pattern recognition and prediction. As an objective weighting method, the entropy weight method can avoid the interference of human factors in the weights of each evaluation index, making the evaluation results more realistic. It overcomes the problem that the weighting process of indicators in the current evaluation method is greatly affected by human factors. The present invention uses the entropy weight method and BP neural network to construct a low-cost flower sea plant landscape plant screening system, which can automatically output whether it is a low-cost variety by inputting variety information.

[0115] Embodiment 2: This embodiment provides a low-cost flower sea plant screening system, including:

[0116] A data acquisition module, used to obtain indicators for evaluating low-cost flower sea plants;

[0117] The first data set construction module is used to determine the weight of each indicator and construct a comprehensive evaluation data set;

[0118] The first model training module is used to train a low-cost flower sea plant screening model using a BP neural network framework based on the weights of each indicator and a comprehensive evaluation data set;

[0119] The evaluation module is used to use the information of flower plant varieties in the area to be screened as the input of the trained low-cost flower plant screening model, output low-cost flower plants that meet the requirements, and form a flower plant variety library in the area.

[0120] Embodiment 3: This embodiment provides a method for generating a landscape scheme, such as Figure 2 As shown, the following steps are included:

[0121] S5. Based on the flower sea plant variety library of the area to be generated for the landscape plan obtained by the low-cost flower sea plant screening method, take photos of individual plants to form a low-cost flower sea plant library for the area;

[0122] S6. Construct a flower sea plant landscape effect evaluation system, collect flower sea plant landscape CAD plan drawings, use expert scoring method, score the collected plan drawings according to the evaluation system, obtain the average score as the final score, and construct a flower sea landscape configuration training set; specifically include:

[0123] S61. Construct a flower sea plant landscape effect evaluation system. Based on the literature, the key elements of flower sea plant landscape planning and design were selected. After investigating the degree of attention paid by the public and experts, as shown in Table 2, ecological beauty, formal beauty, and spatial beauty were finally determined as the criterion layer of evaluation indicators.

[0124] Table 2 Evaluation system of low flower sea plant landscape effect

[0125]

[0126]

[0127] S62, expert scoring. Collect the CAD plan drawings of the flower sea plant landscape through network crawling. Use experts to score the database configuration effect according to the standard, get the average score as the final score, and construct the flower sea landscape configuration training set.

[0128] S7, using the training set, inputting the plan and site information into the convolutional neural network, training the model, and obtaining a landscape plan generation model;

[0129] S8. Use particle swarm algorithm to output coordinates and generate a plan view through the coordinate points and plant picture information in the library.

[0130] Convolutional neural networks (CNNs) have shown strong capabilities in the field of image processing. CNN can extract features such as plant type, morphology, color, and spatial relationship between them through its powerful feature extraction and learning capabilities, and automatically generate plant landscape planes through multi-layer convolution and pooling operations. However, the selection and extraction of CNN features often depend on the setting of network structure and parameters. The particle swarm algorithm can indirectly affect the feature selection and extraction process by optimizing the parameters of certain layers or components of CNN (such as convolution layer, pooling layer, etc.). By adjusting these parameters, the particle swarm algorithm can help CNN better capture the key features of plant landscape. The present invention combines particle swarm algorithm and convolutional neural network to automatically layout a model for low-cost flower sea plant landscape configuration scheme. In order to improve the efficiency of low-cost flower sea plant landscape planning and construction.

[0131] The specific method of step S8 is:

[0132] S81. Collect basic information about the plant configuration site, including the size, shape and design objectives of the site; collect environmental information, including the climate and soil conditions of the site;

[0133] S82, taking the collected basic information and environmental information of the plant configuration site and the low-cost flower sea plant library in the area as input, setting the dimension of the particle swarm algorithm to the number of variety libraries*3, and outputting the coordinates (x, y) of the center point of the picture and the number of selected plants;

[0134] When the number is 0, the corresponding coordinate is (-1, -1), indicating that the plant is not selected; when the number is non-zero, the coordinate is the position of the plant, and the set of coordinate points of the plant image is obtained;

[0135] S83. Generate a plan using coordinate points and plant image information, input the plan and site information into a landscape plan generation model, set the number of iterations, and repeat steps S81 to S82 if the number of iterations is lower than the number of iterations. Otherwise, output the plan with the best score as the final result.

[0136] Embodiment 4: This embodiment provides a low-cost flower sea plant screening system, comprising:

[0137] A library generation module is used to obtain a library of flower sea plant varieties in a landscape plan area to be generated according to any one of the low-cost flower sea plant screening methods 1-6, take photos of individual plants, and form a low-cost flower sea plant library in the area;

[0138] The second data set construction module is used to construct a flower sea plant landscape effect evaluation system and collect flower sea plant landscape CAD plan drawings. Expert scoring is used to score the collected plan drawing effects according to the evaluation system, and the average score is obtained as the final score to construct a flower sea landscape configuration training set.

[0139] The second model training module is used to use the training set to input the plan and site information into the convolutional neural network, train the model, and obtain a landscape plan generation model;

[0140] The landscape plan generation module is used to output coordinates using a particle swarm algorithm and generate a plan view through coordinate points and plant image information in the gallery.

[0141] Example 5: Based on Examples 3 and 4, this example supplements the steps of public participation and subsequent related steps, including:

[0142] Research and needs analysis Understand the population, economy, culture and environment of the community where the site is located, organize focus group meetings, collect residents' needs and opinions on the low-cost flower sea landscape plan generated by computer, and identify areas that need improvement;

[0143] Planning and design: Develop and adjust environmental design plans based on public input and feedback, taking into account the actual conditions of the region, resource constraints and feasibility;

[0144] Implementation and monitoring: Based on the final low-cost flower sea landscape configuration plan, organize implementation work and monitor the progress and effects of the project, communicate and exchange with the public on a regular basis, and promptly resolve problems and improve measures.

[0145] As described above, although the present invention has been shown and described with reference to specific preferred embodiments, it should not be construed as limiting the present invention itself. Various changes in form and details may be made without departing from the spirit and scope of the present invention as defined in the appended claims.

Claims

1. A low-cost method for screening flower plants, characterized in that: The steps include: S1. Obtain indicators for evaluating low-cost flower sea plants; S2. Determine the weight of each indicator and construct a comprehensive evaluation data set; S3. Based on the weights of each indicator and the comprehensive evaluation data set, a BP neural network framework is used to train and obtain a low-cost flower sea plant screening model; S4. Use the information of flower plant varieties in the area to be screened as the input of the trained low-cost flower plant screening model, output low-cost flower plants that meet the requirements, and form a flower plant variety library in the area.

2. The low-cost flower sea plant screening method according to claim 1, characterized in that: The specific method for obtaining the indicators for evaluating low-cost flower sea plants in step S1 is as follows: Based on the literature, the key factors for evaluating low-cost flower plants were determined. Then, through public and expert surveys, the main indicators of concern under different criteria were determined to form an indicator system for evaluating low-cost plants, including: (1) The main indicators of concern under the low maintenance criteria are maintenance needs, replacement frequency and economy; (2) The main indicators of interest under the ornamental criteria are the length of the ornamental period, flower color, plant shape, and leaf shape and color; (3) The main indicators of concern under the adaptability criterion are tolerance to barrenness, shade tolerance, drought tolerance, waterlogging tolerance, disease and pest resistance, reproductive capacity and locality.

3. The low-cost flower sea plant screening method according to claim 1, characterized in that: Step S2 determines the weight of each indicator and constructs a comprehensive evaluation data set, including: S21. Determine the weight of each indicator using the entropy weight method; S22. Construct a comprehensive evaluation dataset.

4. The low-cost flower sea plant screening method according to claim 3, characterized in that: The entropy weight method described in step S21 is used to determine the weight of each indicator. The specific method is: S211, let e j is the entropy value of the jth evaluation index, then the entropy value e j The calculation process is: In the formula, f ij is the characteristic weight of the i-th system under the j-th index; x ij is the observed data of the jth indicator in the i-th system, i=1,2,…,n; j=1,2,…,m; is the sum of all system observation data of the jth indicator; S212, let w j * is the entropy weight of the jth evaluation index, then the entropy weight of the index is as follows: In the formula, e i is the entropy value of the i-th indicator; S213, let P i is the evaluation score of the ith plant. According to the linear weighted comprehensive evaluation formula, the comprehensive evaluation score is: In the formula, p ij is the normalized score of the evaluation index; W j * is the weight of the jth evaluation index.

5. The low-cost flower sea plant screening method according to claim 3, characterized in that: The specific method of constructing a comprehensive evaluation data set in step S22 is as follows: Through a questionnaire survey, landscape design practitioners were asked to score each indicator from 1 to 10 points. Qualified questionnaires were retained as the original sample data, and the training set and test set for training the BP neural network model were constructed in a ratio of 9:

1.

6. The low-cost flower sea plant screening method according to claim 1, characterized in that: In step S3, according to the weights of each indicator and the comprehensive evaluation data set, a BP neural network framework is used to train and obtain a low-cost flower sea plant screening model. The specific method is: S31. The low-cost plant evaluation index is used as the input layer of the BP neural network, and the low-cost flower sea plant evaluation score is used as the output layer of the BP neural network. The number of hidden nodes of the BP neural network is determined according to the trial and error method and the formula, thereby establishing the topological structure of the BP neural network: The number of input, hidden and output nodes of the BP neural network are N, L and M respectively, and the input vector is X = [x0, x1, ..., x N-1 ], the weights between hidden layer point j and input layer point i and output layer node k are V ij and W jk , the thresholds of the output layer and the hidden layer are θ k and φ j , the output calculation formula of the hidden layer nodes and the output layer is: Calculate y k With the target output d k Deviation: δ k (d) k -and k )and k (1-and k ); The weights are resized to: ΔW jk (n)=ηδ k h j ; The weight adjustment formula is: W jk (n+1)=W jk (n)+ΔW jk (n)+μΔW jk (n-1); V ij (n+1)=V ij (n)+ΔV ij (n)+μΔV ij (n-1); S32. The evaluation levels are divided and the comprehensive evaluation scores are judged. The comprehensive evaluation scores ≥70% are classified as Class I, and the comprehensive evaluation scores <70% are classified as Class II. Class I plants are defined as low-cost plants; S33. Use the comprehensive evaluation data set for training, construct a mapping from indicators to evaluation levels, directly output whether it is a low-cost flower plant, and obtain a low-cost flower plant screening model.

7. A low-cost flower sea plant screening system, characterized in that: include: A data acquisition module, used to obtain indicators for evaluating low-cost flower sea plants; The first data set construction module is used to determine the weight of each indicator and construct a comprehensive evaluation data set; The first model training module is used to train a low-cost flower sea plant screening model using a BP neural network framework based on the weights of each indicator and a comprehensive evaluation data set; The evaluation module is used to use the information of flower plant varieties in the area to be screened as the input of the trained low-cost flower plant screening model, output low-cost flower plants that meet the requirements, and form a flower plant variety library in the area.

8. A method for generating a landscape scheme, characterized in that: The steps include: S5. Based on the flower sea plant variety library of the area to be generated for the landscape plan obtained by the low-cost flower sea plant screening method according to any one of claims 1 to 6, take photos of individual plants to form a low-cost flower sea plant library for the area; S6. Construct a flower sea plant landscape effect evaluation system, collect flower sea plant landscape CAD plan drawings, use expert scoring method, score the collected plan drawing effects according to the evaluation system, obtain the average score as the final score, and construct a flower sea landscape configuration training set; S7, using the training set, inputting the plan and site information into the convolutional neural network, training the model, and obtaining a landscape plan generation model; S8. Use particle swarm algorithm to output coordinates and generate a plan view through the coordinate points and plant picture information in the library.

9. The method for generating a landscape scheme according to claim 8, characterized in that: In step S8, the particle swarm algorithm is used to output coordinates, and a plan view is generated through the coordinate points and the plant picture information in the library. The specific method is: S81. Collect basic information about the plant configuration site, including the size, shape and design objectives of the site; collect environmental information, including the climate and soil conditions of the site; S82, taking the collected basic information and environmental information of the plant configuration site and the low-cost flower sea plant library in the area as input, setting the dimension of the particle swarm algorithm to the number of variety libraries*3, and outputting the coordinates (x, y) of the center point of the picture and the number of selected plants; When the quantity is 0, the corresponding coordinate is (-1, -1), indicating that the plant is not selected; When the number is non-zero, the coordinates are the location of the plant, and the set of coordinate points of the plant image is obtained; S83. Generate a plan using coordinate points and plant image information, input the plan and site information into a landscape plan generation model, set the number of iterations, and repeat steps S81 to S82 if the number of iterations is lower than the number of iterations. Otherwise, output the plan with the best score as the final result.

10. A landscape scheme generation system, characterized in that: include: A library generation module is used to obtain a library of flower sea plant varieties in a landscape plan area to be generated according to any one of the low-cost flower sea plant screening methods 1-6, take photos of individual plants, and form a low-cost flower sea plant library in the area; The second data set construction module is used to construct a flower sea plant landscape effect evaluation system and collect flower sea plant landscape CAD plan drawings. Expert scoring is used to score the collected plan drawing effects according to the evaluation system, and the average score is obtained as the final score to construct a flower sea landscape configuration training set. The second model training module is used to use the training set to input the plan and site information into the convolutional neural network, train the model, and obtain a landscape plan generation model; The landscape plan generation module is used to output coordinates using a particle swarm algorithm and generate a plan view through coordinate points and plant image information in the gallery.

Images