4.2. Index Set for Monitoring Sectors’ Digital Transformation
In this section, the indexes for monitoring sectors’ DT are designed from four aspects, i.e., single-domain digitalization, integration and interconnection, collaboration and interaction, and mode innovation.
4.2.1. Indexes of Single-Domain Digitalization
The indexes of single-domain digitalization are as follows:
(a) Digital rate of production equipment, which refers to the ratio of the quantity of digitalized production equipment to the quantity of all production equipment. Specifically, in process industry, digitalized production equipment means independent equipment which can collect information automatically. In discrete industry, digitalized production equipment means NC machines, CNC machine centers, industrial robots, electromechanical equipment with data interface and so on.
(b) Digital R&D tool application rate: Digital R&D design tools mean software tools that are able to help enterprises in product design. These tools enable the enterprises to fulfill many tasks in digital form, such as modeling, simulation, verification and so on. In discrete industries, it refers to the ratio of the quantity of enterprises which can carry out 2D or 3D digital modeling to the quantity of all enterprises. In process industries, it refers to the ratio of the quantity of enterprises that use digital methods in R&D design.
(c) Numerical control (NC) rate of key processes: In discrete industries, it refers to the coverage rate of numerical control systems in key processes. NC DNC and CNC FMC are both typical numerical control systems. In process industries, it means the coverage rate of process control systems in key processes. PLC, DCS, and PCS are all typical process control systems.
(d) The E-commerce application rate refers to the fact that that the transactions of products or services are completed through the Internet. In related enterprises, the payment and delivery of those transactions are completed “online” or “offline”. It should be noted that orders which are received or dispatched by phone, fax or E-mail are not in the scope of E-commerce.
4.2.2. Indexes of Integration and Interconnection
The indexes of integration and interconnection are as follows:
(a) Proportion of enterprises that achieve the integration of design and manufacturing: In the enterprises that achieve the integration of design and manufacturing, data from product design, process design, and product manufacturing are integrated, and the information among those are also transmitted automatically.
(b) Proportion of enterprises that achieve integration of supply production-sale: In the enterprises that achieve integration of supply production-sale, by integrating corresponding information systems, integrated operation between links on internal supply chain can be achieved, so that the management for the procurement of materials, the warehouse of primary materials and completed products, the sales of products, etc. can be seamlessly connected with financial management by collecting automatically data from the integrated information system without manual input.
(c) Proportion of enterprises that achieve integration of management and control: In the enterprises that achieve integration of management and control, corresponding information systems are integrated so that seamless connection is achieved among enterprise production management (plan layer), workshop manufacturing execution (execution layer), production and manufacturing process control (control layer). The upload of information and the send of instructions can go through the above links.
4.2.3. Indexes of Collaboration and Interaction
The indexes of collaboration and interaction are as follows:
(a) Proportion of enterprises that achieve the control of the whole lifecycle of products: In the enterprises that achieve the control of the whole lifecycle of products, all data throughout the product life cycle is managed and used effectively, including product design, technological design, production and manufacturing, field installation, and debugging. Moreover, information can be transmitted automatically not only between product design and process design but also between process design and manufacturing. In addition, unified definitions of digitalized products are applied at all stages of the whole lifecycle of products.
(b) Proportion of enterprises that achieve the collaboration over the industry chain: In the enterprises that achieve the collaboration over the industry chain, the information system is used to achieve the collaborative operation of core businesses such as R&D, procurement, production, sales, finance, etc. with their upstream and downstream enterprises in the industry chain.
(c) Readiness rate of smart manufacturing: In the enterprises that achieve smart manufacturing, the NC rate in key processes exceeds 50 percent, and the integration of supply production-sale has almost completed. These enterprises have high NC degree of underlying equipment, and achieve not only the integration of procurement, production, sales, inventory, and finance on internal supply chain but also that of management and low-level automation. These enterprises tend to be intelligent factory or intelligent enterprises, and have formed the networked, flexible, and intelligent manufacturing mode.
4.2.4. Indexes of Mode Innovation
The indexes of collaboration and interaction are as follows:
(a) Proportion of enterprises that achieve networked collaboration, which refers to the proportion of enterprises that have achieved networked collaboration among all above-scale discrete manufacturing enterprises. In our work, networked collaboration involves cross-enterprise networked collaborative product design, cross-enterprise networked manufacturing and so on.
(b) The proportion of enterprises that achieve service-oriented manufacturing, which refers to the proportion of enterprises that have achieved service-oriented manufacturing among all above-scale discrete manufacturing enterprises. In our work, service-oriented manufacturing involves online operation and maintenance, remote monitoring, precise networked marketing, innovative service provision on the basis of intelligent terminals and so on.
(c) Proportion of enterprises that achieve personalized customization, which refers to the proportion of enterprises that have achieved personalized customization among all above-scale discrete manufacturing enterprises. Enterprises achieving personalized customization are capable of arranging and optimizing productions automatically based on customer orders and material supply.
4.3. Weights of Indexes
As shown in
Table 1, the monitoring index system for sectors’ DT includes 13 monitoring indexes. To score the overall level of a sector’s DT, a weight method of interval valued hesitant fuzzy entropy are explored here for weight assignment.
Analytic hierarchy process (AHP) can analyze the problems hierarchically and combines the evaluation indicators according to different levels [
17]. However, the importance of judgment matrix depends entirely on expert experience with large subjective components and relatively large amount of calculation. When the relationship of elements is complex, principal component analysis (PCA) is a good choice [
18]. However, principal component analysis may lose the original information, which will affect the final weight result.
The entropy weight method (EWM) determines weight according to the amount of information [
19]. That is to say, the more comprehensive effect an element can play, the more weight it will obtain. However, the entropy weight method has no fuzziness. In practice, it is sometimes difficult to give an accurate value, so it is difficult to quantify the qualitative problem reasonably.
When scoring the importance of an element, experts may hesitate between several scores. The weight determined by interval hesitant fuzzy entropy can weaken the uncertainty of experts in subjective scoring [
20], and further make the determination of evaluation weight more reasonable [
21].
In this section, we use interval valued hesitant fuzzy entropy (IVHFE) to determine the weight of each index. The proposed monitoring index system can be depicted as a hierarchical structure as in
Figure 3. IVHFE weakens the hesitation of experts’ subjective evaluation and calculates the first-level indexes and the second-level indexes respectively. A fuzzy model of interval optimization can determine the weights of all indexes more reasonably.
The evaluation interval given by experts is an interval hesitant fuzzy element, and the set of these fuzzy elements is an interval fuzzy hesitant set.
If the expert
q gives a scoring interval for index
i, the interval is Z(
q) = [
m,
n]. Then the interval hesitant fuzzy element of index
i is:
where
is interval left value ‘
m’,
is interval right value ‘
n’.
In the process of scoring, different experts have different understanding of the field and the depth of research. Therefore, when calculating the entropy of the interval hesitation fuzzy element, different experts should be given different weights
. The interval hesitant fuzzy element, which is the score given by all experts for indicator
i, is substituted into (2) to calculate the interval hesitant fuzzy entropy:
where
T is the number of experts, Z(
q) is the score interval of the
qth expert,
is the hesitant fuzzy entropy of the index
i. By calculating entropy values of all indexes at the same level,
, the weight of the index
i, can be derived as follows:
where
N is the number of indexes at the same level, and we have:
.
Constructing an expert set
E = {
E1,
E2, …,
Eq, …,
ET}, then interval hesitant fuzzy elements of indexes in the first-level are as shown in
Table 2.
By substituting the interval hesitation fuzzy element to (2), the interval hesitation fuzzy entropy of the first-level indexes is obtained as
. Then by (3), their weight values
can be calculated, see
Table 3.
According to (3), we can obtain:
The weight calculation of the second-level indicators is similar to that of the first-level indicators. The interval hesitation fuzzy element of the second-level indicators is substituted into (2) to obtain the corresponding set of interval hesitation fuzzy entropy. The sets of interval hesitation fuzzy entropy of the indicators in the second level are as follows: . Substituting the calculated interval hesitation fuzzy entropy into (3), the corresponding set of weight values is obtained.
The weight of the first-level indicator Ai to the target problem is recorded as , and the weight of the second-level indicator directly connected to the first-level indicator relative to its upper-level indicator is recorded as .
Then the weight of the second-level indicator
relative to the target problem is
. According to (3), we can obtain
where
t is the number of the second-level indicators corresponding to the first-level indicator
. According to (5), we know that the sum of the weights of indicators on the second level relative to the target problem is 1.
The weights of indicators on the second level relative to the target question are shown in
Table 4.
4.4. Methods and Steps for Calculation
In the monitoring index system for sectors’ DT, the score of every index and the overall level can be calculated as three steps, as shown in
Figure 4.
4.4.1. Data Collected from Sample Enterprises
To monitor sectors’ DT and collect sample data, we developed the Contemporary Service Platform of Integration of Industrialization and Industrialization (
www.cspiii.com accessed on 31 August 2020) [
59]. According the monitoring index system of DT, combined with sector characteristics, a set of questionnaires is designed for data collection from enterprises, which includes numerical questions, single choice questions, and checklist questions.
For numerical questions, the scoring can be calculated by the formula: . Where means the score of this data collection item, which of value is between 0 and 100; means the real value of the ith data collection item; and are the minimum and maximum thresholds of this data collection item respectively.
For single choice questions, the scoring is calculated as follows:
(i) all the available options should be assigned with certain scores, which should be distributed in the value between 0 and 100.
(ii) the score of the data collection items should be the corresponding score of the selected option.
For checklist questions, the scoring is calculated as follows:
(i) all the available options should be assigned with certain scores, which should be distributed in the value range of [0, 100], and the options’ total scores should be 100.
(ii) the score of the data collection items should be the sum of the selected options’ scores.
4.4.2. Score of Each Index of a Sector
The calculation methods for the value of each index are shown in
Table 5. Normalization is used here to transform the index value to the score of the index ranged in [0, 100].
4.4.3. Total Weighted Score of a Sector
The total score of a sector’s DT is derived by performing a weighted summation of all indexes with the weights in
Table 4.
This score can be regarded as an objective evaluation result of the DT level of a sector. By continuously tracking the score of the monitoring index system, it is possible to effectively judge and predict the development degree and trend of the sector’s DT.