Abstract: Background: Evaluation the operation of vessel traffic service. Materials and Methods: Delphi, fuzzy comprehensive evaluation method and information entropy are supplied to comprehensive evaluate 5 VTS centers in Z province respectively, then Kendall is used to test consistency of the results of the three models. The final evaluation result is combined through mean value, Bora, Compeland and fuzzy Borda. Results: The five VTS centers in the Z province are ranked according to their service efficiency. Conclusion: The combination evaluation method reconciles the differences in the results caused by the evaluations by independent methods and provides a useful reference for the comprehensive and scientific evaluation of VTS.
INTRODUCTION
With the development of the shipping industry and the increasing vessel types, quantity and speed accelerate vessel traffic and density, resulting frequent traffic accidents and water traffic safety supervision are confronted with more pressure and the requirements pertaining to dynamic monitoring, data collection, navigation aid and traffic management capacity of Vessel Traffic Service (VTS) centers continue to rise. It is imperative to conduct VTS comprehensive evaluation and fully exploit and develop the functions of VTS.
Existing literature focuses on the evaluation of economic benefits of VTS and the evaluation methods are mostly limited to the cost/benefit analysis provided in the VTS GUIDE by the International Association of Lighthouse Authorities. Lee et al.1 estimates the public value of a VTS facility construction project using the Contingent Valuation (CV) method. Bukhari et al.2 developed RADAR operated intelligent software, which directly gets the required data from RADAR and displays the vessels list based on their degree of collision severity. Oh et al.3 analyzed the statistical near miss data between fishing vessels and non-fishing vessels in the Wando Vessel Traffic Services (VTS) area and assessed the risk of ship collisions. Mou et al.4 evaluated VTS benefits based on a case study of Zhoushan.
A VTS comprehensive evaluation model based on Delphi method, fuzzy comprehensive evaluation method and information entropy was proposed in this study and the evaluation results that passed Kendall conformance test were combined using mean value method, Bora method and fuzzy Borda method to obtain more objective evaluation.
MATERIALS AND METHODS
VTS comprehensive evaluation model based on delphi method
Construction of VTS comprehensive evaluation index system: According to the definition of VTS and user’s requirements on the functions of VTS in current actual operation, VTS comprehensive evaluation in this study focuses on the system operation and management capacity and resource service capacity. System operation and management capacity primarily reflect the operational performance and the strength of the basic public functions provided by the system. Only the strong capacity in this regard can give full play to other application functions designed and developed by the whole system; resource service capacity primarily reflects VTS's capacity in regulation, analysis and sharing of its acquired information resources of guiding and supporting shipping service; the stronger the function is the greater the VTS's effect on shipping management5. After four rounds of consulting with expert and reaching a consensus, a VTS comprehensive evaluation index system was built (Table 1).
Weighting of analytic hierarchy process: Weighting was made according to the following four steps: Building a hierarchical model, expert judgment matrix and consistency test of judgment matrix, calculating index weight of the indices at all levels and calculating weights of the underlying index to the overall objective. Paired comparisons of the importance of each index were made by a number of maritime experts based on their working and practical experience and the relative weight of each index was calculated by computer program written based on the mathematical model of index weight coefficient (Table 2).
Design of ranking rules of VTS comprehensive evaluation: The ranking rules should be designed based on the characteristics of the index factors to be evaluated in the whole operation of the VTS system and combined with the management methods of administrations to determine specific implementation standards and guidelines. Complying with the above ideology, the ranking rules of the set four-level indices were made in this study (Table 3) and administrations can determine the specific implementation rules according to their actual requirement on management.
Implementation of VTS comprehensive evaluation: The whole VTS comprehensive capacity evaluation process is mainly comprised of ranking process and system comprehensive calculation process, in which the latter is implemented by AHP algorithm generally through automatic computer calculation. While the former can be implemented according to the above ranking rules and it will be ranked after being reviewed by VTS authorities and industry experts.
The specific evaluation time and evaluation experts for each VTS will be appointed by senior administrations. The users of each VTS can identify the shortcomings, developing programs for improvement and promote the effective application of VTS based on the annual evaluation scores of VTS service capability.
| Table 1: | VTS comprehensive evaluation index system |
| Table 2: | VTS comprehensive evaluation index weight |
| Table 3: | VTS comprehensive assessment scoring criteria |
VTS comprehensive evaluation model based on fuzzy comprehensive evaluation method: Fuzzy comprehensive evaluation is a method to build multi-level fuzzy subsets based on an overall analysis of various factors affecting fuzzy objects in a fuzzy environment according to the basic theory of fuzzy mathematics, build evaluation sets for all the possible results and establish appropriate membership functions, make quantitative analysis of the affecting factors that have indistinct borders and have problems in quantitative analysis according to the fuzzy indexes and then make comprehensive evaluation for the fuzzy objects according to the fuzzy transformation principle6. Characterized by several affecting factors, complex structures and powerful fuzziness, VTS comprehensive evaluation fuzzy is suitable for the analysis and quantification of VTS fuzziness via fuzzy comprehensive evaluation method.
Determination of evaluation set of VTS comprehensive evaluation: Evaluation set, also known as comment set or evaluation rating is made up of all the evaluation results of fuzzy objects in fuzzy evaluation and can be expressed as: V = {v1,..., vj,..., vn}, where, vj is several possible evaluation results in evaluation, j = 1, 2,..., n.
The objective of fuzzy comprehensive evaluation is to make an overall analysis of all the factors affecting objects, analyze the possible results obtained in evaluation to build evaluation set, make quantitative analysis of affecting factors according to the fuzzy indexes, carry out comprehensive evaluation based on the fuzzy transformation principle and obtain the optimal results from the evaluation ratings. Refinement degree of evaluation scale will affect the accuracy of evaluation results, the higher the degree, the greater the discrimination between individual index for objects and the more accurate the fuzzy evaluation results, but which will lead to more complex and difficult evaluation process. So it is necessary to select a proper evaluation scale. The selection of evaluation scale involves evaluation scale classification and evaluation scale setting6.
As for VTS fuzzy comprehensive evaluation, the percentage system was adopted in this study, so the evaluation sets were divided into 5 scales in VTS fuzzy comprehensive evaluation model, that is, v = {v1, v2,..., v5}, where, the corresponding percentage interval is v1: Perfect, with scores of 90-100, v2 good with scores of 80-89, v3 medium with scores of 70-79, v4 poor with scores of 60-69, v5 bad with scores of 0-59, as shown in Table 4.
| Table 4: | Percentage of VTS evaluation scale |
The fuzzy evaluation vectors of fuzzy comprehensive evaluation are obtained through the determination of evaluation sets. Membership of evaluation objects in evaluation scales is represented by fuzzy vectors to reflect the fuzziness of evaluation.
Determination of fuzzy membership matrix of VTS comprehensive evaluation: Given the five evaluation scales and the mean scores of VTS comprehensive evaluation (Table 4), the membership functions for the comprehensive evaluation index data xij are as follows:
| • | Membership function of the VTS comprehensive grade of "Perfect": |
| • | Membership function of the VTS comprehensive grade of "Good": |
| • | Membership function of the VTS comprehensive grade of "Medium": |
| • | Membership function of the VTS comprehensive grade of "Poor": |
| • | Membership function of the VTS comprehensive grade of "Bad": |
Fuzzy synthesis of VTS comprehensive evaluation: The fuzzy comprehensive evaluation results C of factor set U are obtained by fuzzy synthesis of the fuzzy evaluation membership matrix R and the corresponding weight vector W, that is:
VTS comprehensive evaluation model based on information entropy-unascertained measure evaluation method: Entropy is the probability of variable uncertainty and can be used to indicate the degree of information orderliness. Information entropy is introduced to measure the average size of information in information sources and thereby represent the average degree of the uncertainty of the entire information system and the more orderly the information, the lower the information entropy, the more uncertain the information source and the greater the information entropy. Information entropy can quantify VTS evaluation data that involve many aspects and have a lot of uncertainties and it can provide decision-makers with more useful information.
Establishment of VTS evaluation index system: According to the index system in Table 1, the marks were re-numbered, as shown in Table 5.
Determination of evaluation matrix and selection of evaluation index: When s objects are evaluated through r evaluation indices and when evaluation object vector is T = {t1, t2,..., ts} and evaluation index vector is I = {I1, I2,..., Ir}, then ti = {ti1, ti2,..., tir}, where tij is the evaluation value i = 1, 2,..., s of evaluation object ti against index Ij and the evaluation matrix is:
The entropy value and entropy weight of the index are calculated using the concept formula of information entropy (that is index discrimination) and the indices that have no discrimination for evaluation objects are deleted and the index system is reintegrated after reduction.
Unascertained measure of single index: Unascertained measure of single index is to calculate the measure uijk of each index by identifying the evaluation scale Q of index and single index measure function U (t) and thus obtain the measure spaces matrix (uijk)s×p of index tij. The details are follows:
Setting p evaluation scales q1, q2,..., qp for tij and then the evaluation vector Q = {q1, q2,..., qp} and Q is an ordered vector, that is, qk>qk+1, then:
Classification criterion matrix:
and uijk = u (tij∈qk) is the degree of unascertain of tij that is obtained via unascertained measure model and belongs to level qk; in addition, single index measure uijk should have the following characteristics:
| Nonnegative boundedness | : | 0≤u(tij∈qk)≤1 |
| Additivity | : | u(tij∈U) = 1 |
| Normalization | : |
Determination of index weight: According to the definition of information entropy, the peak value of index Ij is:
| (1) |
where, p is the number of level, uijk is the single index measure; the importance of index Ij can be indicated by Vij, then the weight of Ij is:
| Table 5: | VTS comprehensive evaluation index system |
| (2) |
Where:
Unascertained measure of multi-index: The synthetic unascertained measure:
of object ti can be obtained based on the weight. Multi-index unascertained measure matrix is as follows:
| (3) |
Confidence Identification: If qk>qk+1 in Q = {q1, q2,...,qp} , the calculation of confidence λ (0.2<λ≤1) leads to:
| (4) |
then ti is considered to belong to level qk. Ranking ti and:
| (5) |
As the ranking rule calculation, where nl is set as a value in an arithmetic progression with a difference of -2 and the comparison and ranking analysis of ti are made according to g(ti).
RESULTS
Evaluation application: With the subjects of 5 Vessel Traffic Service (VTS) centers in Z province, the results of VTS evaluation based on three independent methods and ranking of 5 VTS centers are shown in Table 6. Obviously the ranking results differ from three methods and it is prima facie difficult to judge, which is right or wrong, so consistency testing is made using KENDALL-W concord coefficient, m = 3, n = 5, so 3≤m≤20, 3≤n≤7 and:
is required to be calculated and then checked according to the v alue of table W of Kendall coordination coefficient. Then w = 0.9111 and
Evaluation result combination of the evaluation results was made in this study using the mean value, Bora method, Compeland method and fuzzy Borda method based on three evaluation methods and the combination evaluation results were calculated, as shown in Table 7.
| Table 6: | Results of VTS evaluation and ranking based on Delphi, fuzzy comprehensive evaluation method and information entropy, respectively |
| Table 7: | Combination evaluation results |
DISCUSSION
So far, a few academic studies have been made on the comprehensive evaluation of VTS and there are very few quantitative evaluation methods, including the commonly used Delphi method, analytic hierarchy process and fuzzy comprehensive evaluation method. Due to varied principles and evaluation properties of single evaluation methods, the evaluation results of the same object vary with evaluation methods. In order to reconcile the differences between evaluation results, comprehensive evaluation was made in this study, taking 5 VTS centers in Z province for example, based on the combination evaluation method.
The study of combination evaluation method has been conducted from weight combination, evaluation method combination and combination of single method results. In terms of weight, the subjective weighting methods, such as analytic hierarchy process and fuzzy analytic process, were adopted in this study and moreover, entropy weight information method that uses objective weighting was combined to guarantee scientific and objective target weighting.
So, through the combination of the evaluation results, the rankings of operation effects of 5 VTS centers (A-E) are roughly the same, reconciling the differences in the results caused by the evaluations by independent methods.
CONCLUSION AND RECOMMENDATIONS
A systematic evaluation of the effectiveness of the operation and the implementation of software and management of VTS in Z province was made in this study and an in-depth evaluation was made from VTS structure, function, construction operation and personnel supervision model and combined with many factors, such as personnel and management mode. Study results show that the combination evaluation method is practical and suitable for VTS comprehensive evaluation and provides methods and guidance for the comprehensive and scientific evaluation of the comprehensive operation effects of VTS and can be used as a reference for VTS construction and development.
Specific recommendations include the following two aspects:
| • | With regard to the final evaluation results, VTS (B) present the best operation effect, followed by VTS (D) and (E) and then VTS (A) and (C). When hardware facilities are roughly the same, VTS operation effects depend on the innovation and management capacity of marine attendants and leadership of marine centers, so VTS centers should attach importance to the training and improvement of personnel quality |
| • | There is a space for development and improvement of the comprehensive capacity of VTS centers, which is mainly manifested in the development and exploitation of VTS function application and value-added functions. For instance, there is no available advanced anchor aid inductive analysis or berth aid inductive analysis function in any VTS centers and they fail to make full use of the application of existing data analysis features to navigation decision analyses. Therefore, importance should be attached to the development of value-added services of the existing VTS data |