Fuzzy Comprehensive Evaluation Model of Project Investment Risk Based on Computer Vision Technology

,


Background and Signifcance.
Risk evaluation is an important prerequisite and necessary guarantee for project investment research. Currently, the most popular investment projects include high-quality green food projects, safe meat food production, nursing homes for the elderly, multifunctional cinemas, and various fnancial bonds [1]. However, for most ordinary salary groups, efective communication protocol analysis and design and project investment are mysterious and remote. Limited salary income and unfamiliarity with project investment make these people discouraged from investing in projects that symbolize high risks and high returns. Terefore, research on risk evaluation methods for project investment is of great signifcance. According to diferent salary income groups, the meaning of project investment risks is diferent, and diferent groups have diferent tolerance and acceptance of investment risks [2,3]. Any project from being proposed to formal development to operation and maintenance is full of various uncontrollable factors, which mainly include natural disasters, changes in national policies, and unexpected accidents, which may cause the project to fail to achieve the fnal expected results or benefts, which also leads to investors who invest in these projects. Interests may not grow or even be damaged, which is the main source of risk for project investment [4,5]. Te development and operation of the project are a continuous process, so the risk evaluation of the project is also a dynamic and instant process, so as to help investors accurately and timely understand the risks of the project, so as to obtain greater benefts or encounter a crisis in the project stop loss in time. Computer vision technology is a technology that computer simulates human visual process and has the ability to feel the environment and human visual function. Traditional project investment risk evaluation is mostly based on a large amount of data analysis and related project investment experience to evaluate risks and returns. Although this can generally get a good risk evaluation result, but for natural disasters, policy changes, accidents, and other random accidents, the risks caused by uncontrollable factors often cannot be accurately evaluated and forecasted in time [6]. Computer vision technology can make up for this shortcoming, real-time embedded system, and has important research value for project investment risk evaluation.

Related Research at Home and Abroad.
In terms of project investment risk evaluation methods, related practitioners and authoritative experts have conducted a lot of research as early as the birth of the fnancial management form of project investment. One of the most famous is stock trading. In the 1990s, Wall Street in the United States had set of several stock trading storms. In China, there are also many research results on the evaluation methods of project investment risks. For example, Chao et al. combined modern intelligent transportation projects and related risk assessment and management results, and through expert analysis and fuzzy analytic hierarchy process, they established a risk identifcation and evaluation index system [7]. Tis indicator system can scientifcally identify and evaluate the risks in the project and take corresponding countermeasures. Te only drawback is that it is not widely used because of low acceptance by the masses. Regarding the relatively tense SCP situation in a gas feld in southwest China, Zeng et al. proposed a fuzzy integrated evaluation model for SCP risk evaluation. Te model uses the Delphi method to determine the index weights and establishes the membership matrix based on trapezoidal distribution membership functions and gas feld test data. Tis evaluation model is then used to derive the risk values of the 27 SCP wells in the XX gas feld and the risk degree [8]. However, this method does not incorporate smart sensors for real-time monitoring, and the research is still in experimentation and has not been applied. Wang and Niu evaluated wind power projects through AHP and FCE method to fnd out the actual situation and predict the deviation between the target and the frst-class level. Tis method refects the relationship between the two by ANP and weakens the error caused by independent calculation, which more perfectly solves the shortcomings of point estimation [9]. Su et al. adopted the comprehensive evaluation method of debris fow risk based on fuzzy inference to establish a comprehensive evaluation model of debris fow risk. Tis model provides a debris fow risk evaluation index system for describing various infuencing factors of debris fow risk in hydropower projects [10]. Unfortunately, due to the uncertainty of debris fow in hydropower projects, the risk assessment model still has considerable errors.
For the evaluation of project investment risk, this article adopts fuzzy comprehensive evaluation method combined with computer vision technology in the era of big data. Te fuzzy integrated evaluation method is often used in various risk assessments and value assessments. Jane, a foreign scholar, once introduced risk analysis, network analysis, and gray fuzzy theory to the uncertainty and complexity of largescale engineering projects [11]. Experiments prove that the method is scientifc and efective, but the accuracy of the evaluation results is not high. Voorbis studied the investment information of the Canadian government's national infrastructure projects and found that even though the Canadian government increased national infrastructure spending by 11% from 2015 to 2017 and launched the Canadian Infrastructure Bank to attract private sector project funds, but private investment fell by 18% during the same period [12]. Tis is mainly due to the private investment risks brought by the uncertainty in the project supervision environment, especially oil and gas pipeline projects. James and Vaaler indicated that research in management and related felds is based on the assumption that state ownership of a business increases the risk to private coinvestors. He says the state is not in control but has more equity as well as ownership, which can help the state maintain favorable initial investment project terms for private coinvestors, but likewise the state's ability to intervene in project management under the same initial terms also takes a hit [13]. However, this study did not propose strong measures for this risk assessment and prediction. Frank believes that the management of risk behavior, the consideration of utility, and the tendency to accept certain risks belong to a wide range of behavior and cognitive decision-making. He presents a view of the probability of risk occurrence and the relationship between risk occurrences based on a case study approach [14]. Te results show that the linear relationship between the "probability" of the risk and the "infuence" that produces "value-at-risk" may not be "considered" to be correct, and this relationship may actually be afected by the index.

Innovations in Tis
Article. Tis paper proposes a fuzzy comprehensive evaluation model combined with real-time embedded system and computer vision technology to evaluate the project investment risk, and there may be complex factors such as changes in national policies, weekly transfer of investment funds, and increased risks and costs of investment projects. Based on the diverse types of project investment and the complex factors that cause project investment risks, itt is possible to start from the content analysis and evaluation principles of the present-day balanced evaluation of project investment risk so as to be able to create a three-level project investment risk evaluation index system and give evaluation by the fuzzy comprehensive evaluation method [15,16]. Trough the quantifcation process, scientifc evaluation of the investment risks of existing projects is carried out to provide reference evaluation, coping strategies, and control measures for the investment risks of the majority of project investors and large investment companies to support the development of various investment projects [17,18]. Te article combines the advantages of the traditional AHP and the FCE method; the advantages are that the investment risk is relatively small; reasonable reference can be provided to investors and early warning of investment risk can be carried out. Ten, it evaluates the risk of project investment based on the sources of each risk factor and real-time embedded system and fnally sets the risk level of the investment project through the principle of maximum membership. Te results fnally showed that this method is more accurate than other methods.  [19]. According to the abovementioned risk factor indicators and the principle of AHP, a project risk evaluation hierarchy model can be established, which is mainly divided into target layer, criterion layer, and method layer. Te criterion level is the risk factor of each project, which can be divided into multiple criterion levels according to the importance of the index, the target level is the project information to be invested, and the method level is the method for risk evaluation of the target project. Based on the idea of the FCE method, it is necessary to determine the relevant evaluation index system before establishing the risk evaluation model. As shown in formula (1), this article begins by establishing the risk factor index set I and the secondary risk index set I j under the set I.

Fuzzy Comprehensive Evaluation
In the above formula, I jk represents m the second layer risk under index j, the frst level of risk indices. Te frst-level risk index mainly includes politics, economy, management, organization, technology, and operation. Terefore, the number of frst-level indicators selected in this paper is 6.
According to the abovementioned analysis, the index weight set W is established.
As shown in formula (2), the most important step of the project risk evaluation model is determination and calculation of weights. Tis article is based on the hierarchical analysis method and the entropy value method so as to derive the weights of each risk indicator [20]. Similar to the risk indicator set, weight indicator set W ij represents the secondary risk indicator with weight j and the primary risk indicator with weight i. In addition, it is also necessary to verify whether the determined index meets the sum of the weights of all indices as 1. For the weight group is determined after, it is the calculation of index rating set E and member analysis sets M.
As shown in formula (3), the risk level expressed by each element in the indicator evaluation set is, in descending order, low risk, medium-low risk, medium-high risk, and and risk. Also, the membership degree set corresponding to the elements of these evaluation sets is M.

Constructing a Fuzzy Comprehensive Evaluation Model of
Project Investment Risk. Te diversity of project investment types and the complexity of risk factors determine the difculty of using a single standardized metric to assess its investment risk. Tis paper proposes a fuzzy, comprehensive evaluation method of project investment risk based on computer vision technology. First, the project investment risk evaluation is decomposed into multiple indicators; after that, by analyzing linear algebra such as hierarchical processes and matrix operations, the weights of each level of risk indicators can be derived, and then the importance of each risk factor indicator for diferent levels of project investment can be determined. Finally, we can know the subjective and objective weights of risk factor indicators for each project investment type. Trough the calculation of weights, it is transformed into a set of corresponding evaluation indicators. Te overall risk level of the project investment is then analyzed based on the principle of degree of membership. Te analytic hierarchy process is a systematic, qualitative, and quantitative analysis method, which is applied in many evaluation and judgment models. It combines subjective judgment and objective evaluation to arrive at the fnal decision result, thereby simplifying complex decisionmaking issues.
In this paper, the priority of each level element in the upper level is obtained by using the vector calculation method of solving the characteristic root of the matrix, and fnally, the ultimate weight of the total target is obtained by using the method of weighting and summation. Te purpose of evaluating the weight of project risk factors is based on a specifc, standardized, and quantifed assessment value of the impact of diferent project risk factors and then determining the types of project investment risk factors with high, medium, and low risks. Based on the theoretical basis of the AHP, the project information that investors intend to invest in is viewed as target level A, and the factors that afect investment risks include policy risks, economic risks, management risks, technical risks, organizational risks, and operational risks. We take these factors that afect project investment risk assessment as the basic assessment criterion level B. Te project investment risk level is evaluated by the fuzzy comprehensive evaluation method as plan level C, and then the evaluation model is obtained. After establishing a complete evaluation index system, the corresponding fuzzy judgment matrix can be established. After summarizing and analyzing the project information through the neural network model, scoring the elements of each criterion layer and index layer, the fuzzy judgment matrix J of the frst-level risk factor index can be constructed as follows: As shown in formula (4), m represents the number of frst-level risk factor indicators, n represents the number of scores of the corresponding indicators, and j mn represents the percentage of the number of evaluation times that have obtained the n-item score for the risk level of the m-th indicator factor in the project to the total number of project risk evaluations. According to the judgment matrix of the frst-level risk index, the judgment matrix of the second-level risk factor can also be established, and the subordinate set element value A i of each risk index for the evaluation set E can be obtained by calculation as follows: As shown in formula (5), A i indicates fuzzy comprehensive assessment score of the frst-level risk factor index, where i � 1, . . . , 6; according to the above calculation, member vector A of the corresponding level 1 risk factors can be obtained, while the second-level risk factor index can be further calculated. Te fuzzy comprehensive evaluation set is B.

Establishing the Project Investment Risk Judgment Matrix and Calculating the Weight.
For the hierarchical structure model of fuzzy comprehensive evaluation of project investment risk based on computer vision technology, the degree of importance of the infuencing factors for each evaluation tier in the model is compared with the factors corresponding to the previous layer, so that the relative importance of the infuencing factors in each layer can be known. In this article, the judgment matrix is used to give the determination. Tis type of evaluation model is used to better compare the importance of each element in the standard and target layers. In the article, the importance of each factor in the standard layer is used as a scale from 1 to 9. Te values of the elements in the matrix are written by the given meanings, and thus the judgment matrix is obtained. Also, in the importance evaluation scale, the odd numbered levels from 1 to 9, i.e., 1, 3, 5, 7, and 9, it represents the higher level of importance.
A � As shown in equation (7), formula A indicates the importance judgment value between two risk factor indicators.
Here, a ij represents the judgment value of the importance of the project, that is, risk element a i to risk element a j after mutual comparison. It should be noted that among all elements in the abovementioned judgment matrix, the standard layer importance level evaluation matrix is on the left, M i denotes the product of the elements of the rows in the signifcance measurement matrix, ϖ i denotes the square root of every element in M i in matrix order, and then ω i denotes the worth of every element in ϖ i , which is the ratio of the sum of all elements. M i of each row of the matrix of judgments A is shown in the following equation: Calculation of root values ϖ i of every row in the target matrix is shown in (9), where n indicates determining the order of matrices. Terefore, worth of ω i is possible to obtain the individual elements again in the weight matrix by normalizing the target matrix. 4 Journal of Electrical and Computer Engineering According to the above calculation and analysis, the product of the standard judgment matrix A and the weight matrix ω can be obtained, and several parameter indices of the judgment matrix can be calculated more deeply to verify the identity of the constructed judgment matrix and determine the importance of each infuencing factor. Matrix type calculation formula is shown in the following formula:

Neural Network Prediction Analysis of Project Investment
Risk Level. In the calculation process of the judgment matrix, this paper also verifed the project investment risk fuzzy comprehensive evaluation model through the calculation of the largest eigenvalue λ max of the judgment matrix. According to the average stochastic agreement index questionnaire of the same matrix of ordinal numbers, the stochastic consistency exponent R I � 1.24 of the sixth-order matrix can be obtained, and the consistency index C I and the consistency ratio C R can be further calculated based on the following: By the principle of consistency verifcation, as long as the ratio of consistency is zero, it means that the judgment matrix is completely consistent. Te evaluator can grasp the weights of every indicator in the evaluation index system more easily with his own tendency, by giving appropriate adjustment to its weight and fnally achieving the goal of reasonable weight distribution. Te formula to calculate the consistency index C I and the consistency ratio C R is shown in the following formula: After collecting the main information of the investment project and establishing the judgment matrix, this article divides each risk indicator of the project into three levels for risk evaluation. According to the application of computer vision technology and artifcial intelligence, this article selects the BP neural network (BPNN) in the artifcial neural network, evaluates and analyzes the risk degree of each indicator of the project, and evaluates and predicts the overall risk level of the project. Te algorithm used by the BPNN is the error backpropagation algorithm, which is the BP algorithm. Te BP algorithm can be used in multilayer feed-forward neural networks and also in other felds. It is currently the most successful algorithm. For a BP neural network, suppose the set of random variables is α 1 , α 2 , . . . α ] , α ι represents a node in the network structure, and Φ ] (α t ) represents all the parent nodes of α ι . So, it can be expressed by the following formula: Te joint probability space state is then the product of the conditional probabilities, which are also the probabilities of each variable in a state. After understanding the neural network structure, network learning is limited to parameters. Maximum likelihood and Bayesian approaches are the most commonly used parametric learning approaches that are used in the complete model learning of the dataset. To derive the maximum likelihood function of the parameters α and dataset S from the sample data, one frst needs to determine its network model as follows:

Research Objects.
Te objects of the study in this paper are the fuzzy comprehensive evaluation models based on computer vision technology for the project investment risk evaluation program. According to the fuzzy comprehensive evaluation method, for project investment, risk evaluation is carried out based on the risk factors that all investment projects need to face such as policy, economy, technology, organization, management, and operation. Under the premise of guaranteeing a certain return, investment projects with low risks are given priority for investment, and investment projects with high risks are processed later, which can provide investors with reference data on project investment risks, improve the accuracy and timeliness of project investment risk assessment, and enhance the quality of project investment consulting services. Te main information collection and risk assessment of investment projects are obtained through feld investigation of relevant personnel and summary analysis of expert information to obtain the initial evaluation results, and then the expert evaluation results and real-time project information are fuzzy comprehensively evaluated through the neural network model in artifcial intelligence technology.

Experimental Design.
Tis study collects relevant information of investment projects and evaluation methods for project investment risks through literature research, network investigation, and feld investigation and proposes a fuzzy comprehensive assessment pattern based on project Journal of Electrical and Computer Engineering management risk computer vision technology. Tis experiment is divided into four steps. First, by collecting information and consulting professional investors about the main processes of project investment and risk evaluation methods, the evaluation model for project investment risks is formulated. Ten, we select the members of the investment expert group who will voluntarily participate in the experiment. In the process of selecting the key risk factors of the project, the selected experts should have an in-depth understanding and certain practical experience of the investment project according to the diferent requirements of the project type, so as to ensure the reliability of the results. In addition, this article also selects survey subjects from the project-related frontline practitioners and investors who have invested in the project to conduct a questionnaire survey. Ten, in accordance with the AHP and FCE models, project investment experts and related project practitioners will evaluate project risks from various risk factors and improve the rating index system. Finally, adoption of the maximum membership guidelines and neural network model, the expert's risk evaluation of each risk factor is summarized and analyzed, and a more accurate evaluation result of the overall project investment risk level and investment recommendations are given.

Experimental Data Processing and Error Analysis.
Experimental data handling methods in this paper are mainly realized through the neural network model. Te main principle is an algorithm for error backpropagation. In accordance with the functional characteristics of neural network models and the commonly used experimental data processing methods, this paper uses SPSS22.0 software to help with experimental data. Among them, the most important application of a neural network is the application of data information, which is the transfer function and error analysis function of neurons. Te transfer function selected in this article is the sigmoid function, which is also the activation function of the neural network, and its mathematical model is an S-shaped curve, so it is also called the S growth curve. Its function expression is shown in the following formula: When calculating data through a neural network, the initial thing to determine is the error function. Tere are three types of error functions in neural networks, namely, transfer error, global error, and mean-square error function. It is known from many studies that the mean square error function is a combination of the advantages of the other two types of functions, so the article chooses the mean square error function for error analysis. Te calculation of the mean square error in neural networks is expressed by the following formula:

Investigation on the Status Quo of Project Investment Risk
Evaluation. After understanding the project investment risk evaluation method, this paper proposes a fuzzy comprehensive evaluation model of project investment risk based on computer vision technology and establishes a relatively complete risk factor evaluation index system. Te frst-level risk indicators mainly include policy risk, economic risk, and risk factors such as organizational risk, management risk, technical risk, and operational risk. We select 10 investment experts in this feld to conduct risk assessments, as shown in Table 1. It can be seen from the table that the weight of technical risk in the project investment risk evaluation ranks frst, and the sum of the 10 experts' evaluation levels of technical risk reached 81 points, indicating that the risks brought by technical factors in project investment occupy an important position. Based on the above research, the article chooses the AHP and the fuzzy integrated evaluation methods to give the evaluation the six frst-level risk indicators, and the evaluation model is improved by combining the two methods in this paper. As shown in Figure 1, the evaluation scores of the frst-level indicators of these six project investment risks are all higher than the evaluation results of purely using AHP and FCE methods.
As shown in Figure 1, AHP has the highest score in the operational risk index, 4.6 points, and the lowest score is technical risk, 2.9 points; the FCE method has the highest score of 5.4 in policy risk indicators and the lowest score of 2.8 in organizational risk indicators. Te IFCF method scored the highest in the operational risk index, 6.9 points, and the lowest in the organizational risk index, 5.1 points.

Fuzzy Comprehensive Evaluation Model on the Investment of Chemical Fiber Projects.
After establishing the FCE model of project investment risk, this paper selects some specifc project investment cases for risk evaluation, as shown in Figure 2. Taking the investment situation of chemical fber projects in 2018 as an example, research shows that the investment amount of chemical fber projects in 2018 generally showed a slow growth trend, and its yearon-year growth rate decreased to negative growth in March, and the year-on-year growth rate was relatively stable at other times.

CR Verifcation of Fuzzy Comprehensive Evaluation
Model of Project Investment Risk. To validate the article's evaluation model feasibility, the article uses CR consistency verifcation to test the risk evaluation of the evaluation model of this research for each risk factor index. As shown in Table 2, the main test data include the largest characteristic root of judgment and proof introduced in the abovementioned method, average random consistency index, consistency index and consistency ratio, and index weight.
Tis paper also uses the established project risk fuzzy comprehensive evaluation model to carry out risk assessments on examples of oil and natural gas productionrelated projects. As shown in Figure 3, from 2004 to 2018, the country's natural gas production has grown steadily, but consumption is mainly divided into three types, namely, centralized production and operation, pipeline transportation and sales, and a small number of households for self-production and self-use. It can be seen from the fgure that the proportion of natural gas consumption types for self-production and self-use is increasing year by year. Tis is also one of the important factors to be   Journal of Electrical and Computer Engineering considered in assessing the investment risk of natural gas projects.

Investment Risk Evaluation of Natural Gas in Recent
Years. Because there are variety of project investing types, this paper selects the investment situation of natural gas projects by collecting relevant data and uses the risk evaluation model of this research to conduct risk assessment. As shown in Figure 4, since 1998, with the advancement of oil and natural gas extraction technology and the increase in people's demand for new energy, global natural gas production and consumption have been increasing steadily.

Conclusions
Tis article frst investigates and analyzes the current development status of project investment in this feld, and it was found that most investors have a very simple understanding of project investment risks and most of the   investors blindly follow the trend. It is found that there are defects compared with the methods proposed in this paper, and this paper has more advantages in project investment risk evaluation performance.
Based on this, this article compares and analyzes the traditional project investment risk assessment methods for project investment risk evaluation performance and proposes a risk assessment strategy based on the computer vision technology and real-time embedded system of the FCE model of project investment risk. Tis approach takes advantage of the AHP and fuzzy comprehensive evaluation methods of the traditional value-at-risk evaluation. It can evaluate the risk of project investment in more depth and concretely from all aspects and provide investors with more accurate risk assessment and investment advice. Te development of it is of great signifcance.
To investigate the evaluation performance of the FCE model of the project's investment risk, this paper uses a realtime embedded system and sets up corresponding research experiments for analysis. Te main research results have the following aspects: frst, this article introduces the fuzzy comprehensive evaluation method and risk evaluation principles used in project investment risk identifcation and the steps of evaluation implementation. Second, this article uses a neural network and a real-time embedded system to identify and analyze the risks that investors may face when participating in project investment. On the premise of ensuring a certain return on project investment, a scientifc and systematic risk evaluation index system has been established to evaluate project investment risks from all aspects. Finally, according to the principle of maximum membership degree, the overall risk level of the project is given through a summary analysis of the evaluation indicators at all levels of risk factors.
Because there is not a lot of experience in project investment, the research on project investment risk evaluation in this article is still in the experimental stage. For specifc investment projects, the sources of risk factors are complex and changeable and require more in-depth research. You can also try to combine other risk evaluation methods to build a more scientifc and reasonable project investment risk evaluation model and get more accurate project investment risk evaluation results.

Data Availability
Te data that support the fndings of this study are available from the author upon request.

Conflicts of Interest
Te author declares that there are no conficts of interest regarding the publication of this article.