Research on Intelligent Information Processing and Decision Support Methods in Modern Agricultural and Forestry Economic Management

Since China entered the new era, China’s social and economic strength has been greatly improved, and economic construction has also developed steadily in the process of social progress, and the “three rural” issues with agriculture, rural areas and farmers as the core have always been the focus of attention of the state and the people, reflecting the strengthening of the status of agricultural management in social and economic development, and improving the economic management of agriculture and forestry is the key factor in solving the “three rural” problems [1-3]. Vigorous development of agricultural and forestry economic management is not only to manage one of the three agricultural management, forestry management and economic management, the core concept of its development is to promote the organic integration of the three, to realize the overall planning of the three, is the innovation and development of traditional conceptual thinking [4-5]. The main responsibility of agricultural and forestry economic management is to deeply analyze and explore the development situation of agriculture and forestry, strengthen the application of modern technology, and continuously strengthen the reform and innovation of agricultural and forestry economic management mode oriented to the direction of social advancement, in this process, it is also necessary to always carry out the concept of economic management, so as to realize the innovation and construction of agriculture and forestry management. Strengthening the economic management of agriculture and forestry is to better build agriculture and forestry, so that the agriculture and forestry industry can obtain rapid development, and to ensure that the agriculture and forestry industry can be healthy and stable development [6-8].

Digitization is ubiquitous in modern societies and is gaining traction in agriculture from the perspective of its transformative potential and the pressures it faces [9-10]. Digitization of agriculture and related technologies is considered to be aligned with incremental productivity and transformational changes in agricultural innovation systems. The digitization of agriculture involves the development, adoption, and iteration of digital technologies in the agricultural sector, and is currently most often cited in relation to the adoption of precision agriculture technologies, reduced input costs, and increased yields or productivity, with advances in agricultural science and technology making the role of digital technologies well known [11-12]. Significant changes in agricultural systems have been and will continue to be driven by new digital technologies and multiple advances, including real-time monitoring, big data, the Internet of Things (IoT), machine learning, cloud computing, etc., and a new mode of production, industry formations, and business models will take shape. Each new digital technology should not be considered merely as a separate and discrete technological innovation; behind the transformation in productivity and sustainability is a reconfiguration of the value system that needs to be identified, articulated, and reflected [13-15].

At this stage, agroforestry economic management exists agroforestry economy did not form a strong competitive edge, did not establish a perfect agroforestry economic management system, agroforestry economic management of the serious loss of talent and other aspects of the problem, it is necessary to strengthen the innovation of the agroforestry economic management model, which is the demand of the advancement of knowledge, the social and economic transformation of the needs of the society, but also to ensure the basis of the stable development of the society. Literature [16] reveals the importance of nutrient-efficient crops for modern agriculture, which helps to reduce the pollution of surface water and groundwater resources by intensive fertilization, and can be used as a component of nutrient management in agroforestry and agriculture integrated systems. Literature [17] analyzed the impact of agroforestry development in Pakistan on the country’s economy and identified the problems of agroforestry development in Pakistan in terms of water scarcity, mismanagement, and natural disasters in the context of research in agroforestry related sectors. Literature [18] systematically reviewed the process of agroforestry digital development in Europe as well as the systems and rules related to the sustainable development of agroforestry, revealed that the regulations related to the digital development of agroforestry are in the initial stage, and pointed out that the future of the agricultural data ownership system influences the digital sustainability of agroforestry development. Literature [19] combined the autoregressive distributed lag (ARDL) bounds test method and dynamic ordinary least squares (DOLS) method to analyze the relevant data, and found that agricultural value-added, urbanization, and transitional use of energy caused the degradation of the forest area and thus led to an increase in greenhouse gas emissions, while the use of renewable energy and the expansion of forests are effective in reducing the emission of greenhouse gases. Based on the land use change perspective, [20] used five alternative integrated assessment frameworks with different land use modules, and found that the significant increase in agricultural productivity and the further development of globalized trade have contributed to the natural ecosystem scale, which has contributed to the reduction of GHG emissions from the land production system, and the reduction of food prices. Literature [21] describes the implementation of the Bioeconomy Strategy in Spain, which focuses on agriculture and biotechnology, aiming at the sustainable and efficient production of bioresources, bioenergy from other sources of non-conventional biomass, and the provision of ecological services in rural and coastal areas. Literature [22] explores the current state of green infrastructure development in the Milan region, which leads to an assessment of land use transformations in the region, and also analyzes the impact of policies on green infrastructure development. The current research on the management of agroforestry economy is very broad, and the main research directions include the change of the nature of agroforestry land use, agroforestry green land cover, greenhouse gas emission reduction, biological resources, and water resources management, etc., but the research carried out by the example of the analysis of the research is relatively small.

Digital technology empowers the agroforestry field to effectively promote the development of the agroforestry field, including improving the efficiency and benefits of agroforestry value chain management, promoting the improvement of agroforestry productivity, and facilitating the in-depth discovery of factors affecting the development of agroforestry. Literature [23] describes the benefits of AI technology for agroforestry value chain management, and provides a systematic review of the research field of agroforestry human-computer interaction based on the three fields of intelligent information fusion, intelligent robotics, and trustworthy decision support, which makes a certain contribution to the intelligentization and informatization of agroforestry management. Literature [24] used a questionnaire survey to investigate farmers’ attitudes toward the practice of digital technology in agriculture, and based on the results of the study, it was found that farmers perceived the productivity improvement brought by digital technology and showed great enthusiasm for learning digital technology. Literature [25] used qualitative textual analysis to illustrate the attitudes of the Food and Agriculture Organization (FAO) and the Organization for Economic Cooperation and Development (OECD) towards digital agriculture as a way to maximize food production through technology, and to propose trade-offs between how digital agriculture can affect the future supply of different food systems, regulation, and cultural ecosystem services. Literature [26] analyzes the current status of the practice of IoT technology in the field of agriculture and forestry, and deeply analyzes the characteristics of IoT technology and the underlying logic of the technology, as well as the path of integration into the development of the field of agriculture and forestry, and combines the relevant literature with examples of discussion. Literature [27] assessed soil quality in all aspects of deforestation and intensive cultivation lands in Mazandaran province, Iran, through digital soil mapping methods, and the results of the study showed that changes in the nature of land use have reduced soil quality, and there is a need to focus on the sustainable use of agricultural land in order to improve soil quality. Literature [28] conducted an expert interview meeting on digital agroforestry development and learned that economic costs and policy regulation are the most notable elements driving the digitalization of agroforestry.

In this research paper, a system designed to assist in decision support for the agroforestry economy was constructed. The system employs a convolutional neural network and an optimized coordinated attention mechanism module (MA) as its core tools for intelligent information processing. In addition, the system integrates a variety of algorithms, such as fuzzy hierarchical analysis and entropy weighting, to achieve comprehensive assessment and judgment of agroforestry economic decisions. It is applied to a local agroforestry management department to evaluate the scientificity and practicability of this paper’s system in four dimensions, namely, system operation, planting structure optimization decision-making, planting yield prediction of agroforestry crops, and weighted analysis of agroforestry projects.

2

Intelligent information-processing-based yield prediction techniques for agricultural and forestry crops

2.1

Convolutional Neural Networks

CNN is a deep learning model or a multilayer perceptron similar to an artificial neural network, which is commonly used to analyze visual images [29] and the convolutional neural network architecture is shown in Figure 1.

The convolutional neural network contains a data input layer, a convolutional computation layer, a non-linear layer, a pooling layer, and a fully connected layer. The main function of the data input layer is to preprocess the initial data collected. First, all dimensions of the input data are adjusted to be centered at 0, which means that the center of the sample is returned to the origin of the coordinate system. Secondly, the magnitude is unified in the same range by a normalization operation, so as to reduce the interference caused by differences in the range of values between different dimensions. PCA dimensionality reduction techniques and whitening methods can also be applied to the data. The convolutional computational layer simulates the receptive field in a biological visual system, where features are extracted from the input image using a filter. Assuming an input image and a filter, the convolution operation can be expressed as: (1) $Y = X * W$

Where, Y is the feature map after convolution and ^* denotes the convolution operation.

The mathematical expression for the convolution operation is: (2) $(X * W) [n, m] = \sum_{i = 0}^{H - 1} \sum_{j = 0}^{W - 1} X [i, j] \cdot [n \cdot s t r i d e + i, m \cdot s t r i d e + j]$

where n and m are positions corresponding to those in the feature map Y. H and W are the height and width of the input image X. stride is the step size, which indicates the number of pixels moved by the filter after each convolution operation.

The nonlinear layer can also be referred to as the activation layer.The ReLU activation function is the most commonly used nonlinear activation function and is given by: (3) $f (x) = m a x (0, x)$

The pooling layer is used to reduce the size of the feature map while retaining important information. Maximum pooling can be expressed as: (4) $Y = m a x (X, p o o l s i z e)$ $$Y = max(X,pool\,size)$$

where pool size is usually a small window such as 2 × 2 or 3× 3 . Mean pooling can be expressed as: (5) $Y = \frac{1}{p o o l s i z e^{2}} \sum_{i = 0}^{p o o l s i z e - 1} \sum_{j = 0}^{p o o l s i z e - 1} X [i, j]$

A fully connected layer is a neural network structure that establishes connections between each neuron in the current layer and all neurons in the previous layer. Its main function is to learn complex associative relationships between input features.

2.2

Attention Characterization Enhancement Mechanisms

By improved coordinated attention mechanism module has plug and play properties and lightweight structure [30-31]. The structure of the improved coordinated attention mechanism used in this paper is shown in Figure 2.

The module first calculates the mean value of the feature map in x, y coordinate on each channel, i.e., the two spatial ranges (H,1) and (1,W) of the pooling kernel are utilized to encode each channel along the horizontal and vertical coordinates, respectively, using the following equations: (6) $z_{c}^{h} = \frac{1}{W} \sum_{i = 0}^{W} x_{c} (h, i)$ (7) $z_{c}^{w} = \frac{1}{H} \sum_{j = 0}^{H} x_{c} (j, w)$

where $z_{c}^{h}$ and $z_{c}^{w}$ denote the average features done on the feature map elements from height and width directions, W denotes the feature map width, H denotes the feature map height, and x_c(h, i) and x_c(j, w) denote the feature vectors in horizontal and vertical directions, respectively. The aggregated information in the height and width directions is computed by the above two formulas to produce a pair of orientation-aware feature maps $z_{c}^{h}$ and $z_{c}^{w}$ . These two feature maps containing positional information have a larger capacity than using only the global averaging pool.

The resulting two arrays are then passed in series through a convolutional layer with a kernel size of 1 × 1 to reduce the number of channels, where the transform expression can be written as: (8) $f = δ (F ([z^{h}, z^{w}]))$

Where [z^h, z^w] denotes two average feature maps connected in series along the spatial dimension, δ is the nonlinear activation function, F ∈ R^Cr*(H+W) refers to the intermediate feature maps encoding the space in the horizontal and vertical directions, and r is the feature map downsampling multiplier. Then, the feature maps are divided into two groups according to the original scale, which are denoted as f^h ∈ R^C/r*H and f^w ∈ R^C/r*W, respectively, and then the convolution operation is performed according to the groups to restore the number of channels to the original number of feature map channels. Finally, after passing the Sigmoid activation function, the weights g^h and g^w in the directions of x and y coordinates are generated to reorganize the original feature map in two directions, and the reorganization process is: (9) $g^{h} = σ (F_{h} (f^{h}))$ (10) $g^{w} = σ (F_{w} (f^{w}))$

Therefore, the computational formula for the generation of the final new feature map can be written as: (11) $y_{c} (i, j) = x_{c} (i, j) * g_{c}^{h} (i) * g_{c}^{w} (j)$

3

Decision support algorithms

3.1

Fuzzy Hierarchical Analysis

Fuzzy hierarchical analysis method (FAHP) is a combination of hierarchical analysis and fuzzy set theory, fuzzy judgment matrix can be obtained according to the relevant definitions and calculation formulas in fuzzy mathematics to obtain a fuzzy judgment matrix with consistency [32-33]. FAHP eliminates the need for consistency testing of judgment matrix in the traditional hierarchical analysis method, which speeds up the convergence of calculation results.

3.1.1

Fuzzy mathematics

In classical mathematics, it is believed that every object can be accurately judged as conforming or not conforming to a specific concept, and a distinct distinction can be established between them. However, in practice, the difference between the objective things in the expression of the process of indistinguishability, such as the ambient temperature can be described as “very hot, hot, comfortable, cold, cold,” and so on, these are fuzzy concepts belonging to the ambiguity.

Definition 1 A known domain X , on which to do the mapping F , there are: (12) $\begin{matrix} F : X \to [0, 1], x \in X \\ x \to F (x), F (x) \in [0, 1] \end{matrix}$

Then F is said to be a fuzzy set on X , denoted as F . F (x) is said to be the affiliation function of fuzzy set F .

Definition 1 it can be seen that the range of values of the domain of the classical set eigenfunction is the discrete set {0, 1} and the range of values of the domain of the fuzzy set eigenfunction is the continuous closed interval [0, 1].

Definition 2 Let matrix R = (r_ij)nn, if 0 ≤ r_ij ≤ 1, then the matrix R is said to be a fuzzy matrix.

Definition 3 Let matrix R = (r_ij)nn be a fuzzy matrix and if r_ij + r_ji, = 1, i, j = 1,2,..., n , then matrix R is said to be a fuzzy judgment matrix.

Definition 4 Let matrix R = (r_ij)nn be a fuzzy matrix and if r_ij = r_ik−rjk + 0.5,i, j, k = 1,2…n, then matrix R is said to be additive consistent fuzzy judgment matrix.

Definition 5 Let matrix R = (r_ij)nn be a fuzzy matrix, if a_ii = 0.5 and a_ij + a_ji = 1, then this matrix is said to be a fuzzy complementary matrix.

3.1.2

Determination of the affiliation function

The affiliation function is a quantitative expression of the fuzzy concept F. When using fuzzy mathematics to solve practical problems, the first step is to determine the affiliation function. In general, the determination of the affiliation function is usually determined by using empirical or statistical methods, while it can be combined with authoritative experts to determine after evaluation. In this paper, the fuzzy statistical method is used.

General steps of the fuzzy statistical method: in each trial it is necessary to make an explicit judgment as to whether a definite element x₀ in the domain belongs to a variable set F^*. Note that F^* must be a well-defined set in each trial. In each trial, x₀ is fixed and F^* varies randomly; if element x₀ belongs to F^* as many times as m in n trials, then the frequency of affiliation of element x₀ to F can be defined by the following equation:

The number of times m/total number of experiments n for the affiliation frequency $= \frac{x_{0}}{\in} F^{*}$ of x₀ to F .

As the number of n increases, the frequency of affiliation of element x₀ gradually stabilizes around a certain value, and this stabilized number is the affiliation of x₀ to F .

3.2

Entropy weight method

The core idea of the entropy weighting method is to extract information changes of indicators by calculating the information entropy of each indicator. The algorithm determines the size of the indicator weights by analyzing the relative degree of change of the indicators and their impact on the overall situation.

The entropy weight method of calculating indicator weights mainly includes the following steps: 1)

Dimensionless processing of all indicator data.

According to formula (13), the evaluation indicator data are preprocessed to obtain matrix Z: (13) $Z_{i j} = X_{i j} / \sum_{i = 1}^{m} X_{i j}$

Where m denotes the number of objects to be evaluated in the experiment and n denotes the number of indicators.

2)

Calculate the information entropy value of the indicator H_j.

The specific calculation is shown in Equation (14): (14) $H_{j} = (- \frac{1}{\ln m}) \sum_{i = 1}^{m} f_{i j} \ln f_{i j}, f_{i j} = Z_{i j} / \sum_{i = 1}^{m} Z_{i j}$ $${H_j} = \left( { - {1 \over {\ln \,m}}} \right)\mathop \sum \limits_{i = 1}^m {f_{ij}}\,\ln {f_{ij}},{f_{ij}} = {Z_{ij}}/\mathop \sum \limits_{i = 1}^m {Z_{ij}}$$

3)

Calculate the weight vector W for the final indicator: (15) $W_{j} = (1 - H_{j}) / (n - \sum_{j = 1} H_{j})$

4

Agroforestry economic decision support system design

4.1

Overall System Architecture Design

The agroforestry economic decision support system designed in this paper includes five parts: a database, model library, method library, knowledge base, and interface for human-computer interaction. The overall structure of the agroforestry economic decision support system is shown in Figure 3.

1)

Database

The agroforestry economic decision support system’s data can be categorized as spatial data and attribute data. Spatial data include administrative district maps, topographic maps, water system maps, road maps and thematic maps, etc., which are stored in the spatial database based on the system. Attribute data include statistical data of the agriculture and forestry economic index system, evaluation result data, and other attribute data related to the system, which are stored in the Oracle database.

2)

Model library

The model library consists of two parts: the dictionary library and the file library. The model dictionary library gives a description of the model’s name, number, and file, while the model file library stores the model’s information in the system.

3)

Method library

The method library provides computational methods and algorithm implementation for the system’s thematic model operation. It mainly includes the following methods, hierarchical analysis algorithm, fuzzy judgment matrix algorithm, index weighting algorithm, etc.

4)

Knowledge Base

The knowledge base of this paper stores the more mature experience, knowledge and models of the relevant experts in agricultural and forestry economic research, and according to its nature and characteristics, the knowledge of different classes constitutes a number of sub-knowledge bases.

5)

Human-computer interaction

The human-computer interaction subsystem exists in each functional subsystem rather than separately, and it is integrated into the operation and application of the entire system.

4.2

Overall System Functional Design

The Decision Support System for Agroforestry Economy (DSSAFE) is an integrated information processing system based on geographic information technology (GIT) techniques. The following are the functional modules of this system. Figure 4 illustrates the functional structure of the agroforestry economic decision support system.

1)

Information maintenance

The information maintenance module includes spatial data maintenance and attribute data maintenance. It combines attribute data with geographic information to achieve the unified operation of spatial data and attribute data. And input and edit the attributes through graphics to achieve the economic evaluation of agriculture and forestry.

2)

Economic evaluation of agriculture and forestry

As an important part of intelligent agriculture and forestry, its application also develops towards specialization, socialization, mutual integration, and penetration with other technologies. The functions of this module include model library management, method library management, knowledge library management, evaluation index system management and evaluation result output.

3)

Agricultural and forestry economic decision-making

The module includes production conditions decision-making, production operation decision-making, input-output decision-making and capital flow decision-making. For areas with low average levels, it proposes improvement programs for evaluation indexes with heavy combination weights, and provides managers with auxiliary decision-making suggestions.

4)

Report Management

Report generation will be required based on the user’s query conditions to display the data in tables, which will be intuitively provided to decision makers.

5)

Thematic map management

The system is based on ArcGIS for managing graphics. According to the specified attributes or spatial analysis results to set the graphic color or illustration and other performance object level, load the thematic map title, text, north-south compass and grid, can get all kinds of thematic maps of the economic evaluation of agriculture and forestry.

6)

System Management

User management: mainly used for adding and deleting users, modifying user rights, setting and modifying passwords.

Administrator: staff of agroforestry management department. It can statistically analyze the data of agroforestry economic evaluation indicators and urgent geographic data and update the database.

Advanced users: relevant sections and leading staff of agroforestry management department. They can query and browse the data of agroforestry economic evaluation indicators and urgent geographic outputs, and use the functions of the agroforestry economic evaluation and decision-making module.

Ordinary users: staff of the agroforestry management department. They can query and browse the agroforestry economic evaluation index data and basic geographic data, but they cannot use the data processing function module of the system.

4.3

Agroforestry economic database design

4.3.1

Logical structure of the database

The basic database includes geospatial data and corresponding attribute data, statistical data of agroforestry economic evaluation indicators, etc., and the thematic database includes evaluation result data and generated thematic maps. The logical structure of the database and the main data in the database are shown in Figure 5.

4.3.2

Database creation

1)

Relational database Oracle9i

This paper utilizes object-based relational Oracle 9i to store and manage large amounts of data on the system, ensuring the database’s versatility, portability, and scalability.

2)

Data storage method

For attribute data such as evaluation index data, user data, model index data, etc., Oracle data tables are used for storage. For geospatial data, Geodatabase is used for storage.

3)

Database access

In this system, spatial data is accessed by the ArcSDE spatial data engine, and attribute data is accessed by ADO.

4.3.3

Database updating and management

1)

Database Update

Database update involves adding, deleting, and modifying data in the database. The existing data entered in the database, according to the actual work needs, add or delete the indicators of the economic evaluation of agriculture and forestry, at the same time, found that there are errors in the information entered, the indicator data for the corresponding changes.

2)

Database Management

Database management refers to the configuration of the operating environment of the database, user management, permission allocation and other settings to ensure the consistency and security of the data, to ensure the stability and security of the database, and to reasonably allocate user rights.

5

Effectiveness of the application of decision support systems based on intelligent information processing

5.1

Evaluation of the accuracy of the system’s decision support

In order to verify the correctness of the system’s decision-making and the rationality of the process, this chapter is arranged to simulate and analyze the system. The system is discussed and analyzed mainly from the perspective of the accuracy of the decision-making algorithm and examples of system operations.

Due to the different number of cases of different solutions in the existing case set, the cases are extracted in equal proportion according to the ratio of various solutions in order to ensure the accuracy of the simulation.

Randomly extract 30% of the case data from each of the five different solution case sets as test data, respectively, and combine the remaining 70% of the data into a case base for reasoning.

Taking the problem X to be solved as an example, the maximum similarity distance value of similar cases obtained from the case base in this algorithm simulation is 0.4, so 0.4 is set as the similarity distance threshold for case retrieval. The data in the test set are sorted sequentially according to the method of solving problem X. The solutions included cutting and weeding, full replanting, growth logging, light penetration logging, and integrated nourishment. Comparison with the original solutions in the test set yielded inference simulation results for different solutions, as shown in Figure 6.

The results of the algorithm simulation were analyzed and it was learned that the reasoning accuracy of integrated nourishment was 100%, the reasoning accuracy of comprehensive replanting was 96.3%, and the overall decision-making accuracy was 98.47%. It can be seen that the reasoning accuracy is higher in cases where the technological requirements of the nurturing method are lower, or where there is an intersection with other operational methods. The lower inference accuracy of cases involving full replanting is mainly due to the small amount of data about full replanting in the simulation dataset. Meanwhile, from the technical requirements of comprehensive replanting, it is necessary to utilize historical silvicultural data in addition to the historical data of nurturing, so it is necessary to expand the type and scope of the case base in terms of subsequent algorithm improvement.

5.2

Weighted analysis of the comprehensive evaluation of agroforestry economy projects

In order to verify the accuracy and scientificity of the agroforestry economic decision support system proposed in this paper for project decision-making, in this section of the experiment, it was investigated whether the judgment of the developed system on the project was consistent with the actual situation, i.e., to verify the practicality of the system.

Validation method: first-hand information about the comprehensive agroforestry development project was collected, including the basic information of the project, the feasibility study report, and the acceptance of the project. In this paper, we take the three projects that have been set up in the department of comprehensive agriculture and forestry development as an example, i.e., Project A, B and C. Through the system of this paper, we make decisions on Project A, Project B and Project C, and then compare them with the results of the acceptance, and if the priorities of the three projects are in the same order, then it means that the system developed by using entropy weight optimization and fuzzy hierarchical analysis is scientific and effective.

This study analyzes four dimensions: technical feasibility, market prospects, financial aspects, and organizational management. Each dimension includes several secondary indicators, totaling 22, such as the technical feasibility of integrated agriculture and forestry development projects with technical solutions, equipment solutions, engineering solutions, energy saving measures and water saving measures and other indicators. The scoring results of experts for the 4 dimensions are shown in Figure 7.

As can be seen from the figure, the experts’ overall evaluation of the technical feasibility, market prospect, financial aspects and organizational management of Project B is higher, with the overall rating of 81.7, 88.5, 86.3 and 85.5 in order of magnitude, and Project A, B and C are better than Project A, and Project A is better than Project C in comparison.

The basic information of projects A, B and C is entered into the agroforestry decision support system, and then the projects to be compared are selected in the project selection page, and the project indicators are standardized by the experts in the standardization page of the project indicator evaluation, and the values of the matrix elements are entered in the judgement matrix page to get the judgment results of the system. The system’s comprehensive judgment results for the four dimensions are shown in Figure 8.

The data in the figure show that the judgment results given by the system are consistent with the results of the model, i.e., the comprehensive evaluation of project B in the 4 dimensions is higher and its weight is calculated to be 0.398. It shows that the system better reflects the model in the process of design and development, that is, the decision-making of the system can assist decision-makers scientifically and efficiently to carry out the assessment of agroforestry projects’ weights, and improves the effect of decision-making.

5.3

Yield prediction of agroforestry crops applying CNN-MA

The decision support system for agroforestry economy built in this paper builds a MA model based on convolutional neural network and attention feature increase, in order to verify the excellent performance of these two groups of models when intelligent information processing in decision support system. In this section, the experiment collects the 2018-2023, the unit production of agroforestry economic crops in a certain place, and predicts the unit production of local agroforestry economic crops in 2024 by the system of this paper. The unit production of agroforestry cash crops in 2024 can be predicted using this paper’s system, as shown in Figure 9.

As can be seen from the prediction results of the system, the agroforestry cash crop unit yield maintains the trend of annual growth, and the average annual growth of the local agroforestry cash crop unit yield from 2018 to 2023 is 155.8 kg/year. The system’s predicted yield in this paper is almost identical to the actual yield, and the average prediction accuracy is 99.62%. From this, it can be seen that the convolutional neural network and MA attention mechanism model of this paper’s system performs well in predicting information data. And the system in this paper predicts that the unit yield of agroforestry cash crops in this place will be approximately 9624kg in 2024.

5.4

Decision-making for the analysis and optimization of agroforestry crop cropping structure

The optimization of the planting structure of agroforestry cash crops is a complex process of designing multiple factors, and this section aims to use the decision support system of this paper to provide a place with an optimized structure of the planting area of agroforestry cash crops that improves the efficiency of agroforestry, guarantees food security, and increases the income of farmers. This section focuses on ten agroforestry cash crops, including wheat, corn, rice, soybean, cotton, sugarcane, oilseed rape, coffee, cocoa, and rubber, and provides a suggested planting area for each type of crop by calculating the economic benefits before and after optimization.

Figure 10 shows the ratio of planting area before and after optimization of each type of crop, and the ratio of economic benefits. As can be seen from the figure, the system in this paper suggests that there is an increase and decrease in the planted area for different crops to ensure the maximum economic benefit ratio after structural optimization. The system in this paper predicts that the economic benefit brought by planting area after optimization of each crop is 1.2~1.44 times than that before planning. The agriculture and forestry management departments can use this result as reference data to guide the structural adjustment of agriculture and forestry.

6

Conclusion

In this study, a decision support system for agroforestry economy integrating intelligent information processing was developed by utilizing convolutional neural network and improved coordination attention mechanism module MA, combined with various decision support algorithms. Decision analysis is conducted through the system operation system energy, the optimization of agroforestry crop planting structure, etc., to evaluate the decision-making effect of this system. In this paper, the overall decision-making accuracy of the system is 98.47 when targeting the X problem, which has a good decision-support effect. The system calculates the weight of project B as 0.398, and the evaluation score on technical feasibility, market prospectivity, financial aspects, and organization and management synthesis is 84.57, which is considered that project B is better than projects A and C, which is consistent with the expert evaluation results. Furthermore, the smart information processing model chosen in this paper has an average prediction accuracy of 99.62% for the unit yield of agroforestry cash crops. For different agroforestry cash crops, the system in this paper can provide the planting structure optimization decision with the highest economic efficiency.

Funding:

This research was supported by the Special Doctoral Support upon Completion of Service Period Research Project: Quantitative Study on Forest Ecological Compensation Benefits (Project Number: G2022sk23).

Language:: English

Publication timeframe:: 1 times per year
Journal Subjects:: Life Sciences, Life Sciences, other, Mathematics, Applied Mathematics, General Mathematics, Physics, Physics, other

Journal RSS Feed

Research on Intelligent Information Processing and Decision Support Methods in Modern Agricultural and Forestry Economic Management

Xingang Wang

Published Online: Mar 17, 2025

Received: Nov 16, 2024

Accepted: Feb 18, 2025

DOI: https://doi.org/10.2478/amns-2025-0315

KeywordsIntelligent information processing, Convolutional neural network, Attention mechanism, Fuzzy hierarchical analysis, Decision support

© 2025 Xingang Wang, published by Sciendo

This work is licensed under the Creative Commons Attribution 4.0 International License.

Keywords
Intelligent information processing, Convolutional neural network, Attention mechanism, Fuzzy hierarchical analysis, Decision support