An intelligent site selection method and terminal for inland LNG terminal

Abstract

The present invention relates to the technical field of site selection and evaluation, and specifically discloses an intelligent site selection method and terminal for an inland LNG terminal, the method comprising obtaining a first remote sensing image of a to-be-selected area, determining a refined area based on the first remote sensing image, and obtaining a second remote sensing image in the refined area; extracting a quantitative area and a qualitative area in the first remote sensing image containing the second remote sensing image according to an obtained convolution kernel; identifying the quantitative area to obtain the actual value of the quantitative index; counting the qualitative area, establishing a scoring questionnaire, sending it to a scoring end, and receiving the scores fed back by the scoring end; counting all the scores to obtain a matrix of the to-be-selected area, determining the total score of each sample based on the TOPSIS method, and determining the final to-be-selected area according to the total score. The present invention entrusts the information collection link to the satellite, and after obtaining the remote sensing image collected by the satellite, extracts the information according to the preset index type, extracts only the qualitative index, and hands it over to the expert group for evaluation, which greatly simplifies the workload of the expert group.

Description

Technical Field

[0001] The present invention relates to the technical field of site selection and evaluation, and in particular to an intelligent site selection method and terminal for an inland LNG terminal. Background Art

[0002] LNG terminals are facilities used for loading and unloading liquefied natural gas (LNG), usually located on the coastline. These terminals usually include storage facilities for LNG, unloading equipment, berths, pipeline connections, and related safety systems. At these terminals, LNG is transferred to ships or from ships to land facilities. These facilities play a vital role in the energy industry because LNG is a clean energy source and occupies an important position in international energy trade. Therefore, the design and operation of LNG terminals require strict safety standards and environmental protection measures.

[0003] The existing site selection process of LNG terminals is completely dependent on the expert group, including the information collection and information evaluation stages, which requires a large amount of human resources and an extremely long work cycle. With the development of science and technology, the work process of the expert group can be completely simplified and the efficiency can be improved. How to use intelligent equipment to optimize the site selection of LNG terminals and improve site selection efficiency is the technical problem that the technical solution of the present invention wants to solve. Summary of the invention

[0004] The object of the present invention is to provide an intelligent site selection method and terminal for an inland LNG terminal to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] An intelligent site selection method for an inland LNG terminal, the method comprising:

[0007] Receive the indicator type uploaded by the management party, query the area corresponding to each indicator type, take photos of the area, extract features from the photos of the area, and obtain the convolution kernel of each indicator type; wherein the indicator type includes two parameters, the value of one parameter includes at least guidance and constraint, and the value of the other parameter includes at least qualitative and quantitative; the convolution kernel is a preset image feature;

[0008] Acquire a first remote sensing image of the area to be selected, determine a refined area based on the first remote sensing image, and acquire a second remote sensing image in the refined area; wherein the accuracy of the second remote sensing image is greater than the accuracy of the first remote sensing image;

[0009] Extracting a quantitative area and a qualitative area in the first remote sensing image containing the second remote sensing image according to the obtained convolution kernel; the quantitative area is an area corresponding to the quantitative index in the spliced ​​close-up image, and the qualitative area is an area corresponding to the qualitative index in the spliced ​​close-up image;

[0010] Identify the quantitative area and obtain the actual value of the quantitative indicator; count the qualitative area, establish a scoring questionnaire, send it to the scoring end, and receive the scoring feedback from the scoring end;

[0011] All scores are counted to obtain the candidate area matrix, and the total score of each sample is determined based on the TOPSIS method, and the final candidate area is determined based on the total score.

[0012] As a further solution of the present invention: the steps of obtaining a first remote sensing image of the area to be selected, determining a refined area based on the first remote sensing image, and obtaining a second remote sensing image in the refined area include:

[0013] Acquire a first remote sensing image of the area to be selected, perform contour recognition on the first remote sensing image, and obtain a regional contour;

[0014] When the area of ​​the region outline is included in the preset area range, the region outline is marked as a refined area;

[0015] A second remote sensing image of the refined area is obtained to replace the corresponding area of ​​the first remote sensing image.

[0016] As a further solution of the present invention: the step of extracting the quantitative area and the qualitative area in the first remote sensing image containing the second remote sensing image according to the obtained convolution kernel comprises:

[0017] Read the convolution kernel of each indicator type, traverse the first remote sensing image replaced by the convolution kernel, and calculate the matching degree in real time;

[0018] When the matching degree reaches a preset matching degree threshold, recording the matching position;

[0019] Query the indicator type corresponding to the convolution kernel, and divide the matching position into quantitative area and qualitative area according to the indicator type.

[0020] As a further solution of the present invention: the steps of counting all scores to obtain a matrix of candidate areas, determining the total score of each sample based on the TOPSIS method, and determining the final candidate areas according to the total scores include:

[0021] Arrange each indicator type according to the preset indicator order;

[0022] When the indicator type is a quantitative indicator, read the identified actual value; when the indicator type is a qualitative indicator, read the score as a sample;

[0023] Arrange the samples in the order of the reference time to obtain the matrix of the areas to be selected;

[0024] The candidate area matrices of different candidate areas are spliced ​​together, and the total score of each sample is determined based on the TOPSIS method;

[0025] The candidate area corresponding to the sample with the largest total score is selected as the final candidate area.

[0026] As a further solution of the present invention: the step of splicing the candidate area matrices of different candidate areas and determining the total score of each sample based on the TOPSIS method comprises:

[0027] All data are normalized to obtain a standard matrix;

[0028] Extract the constraint index in the standard matrix and obtain the constraint sub-matrix;

[0029] When any value in the constraint sub-matrix exceeds the preset condition, the sample corresponding to the value is deleted from the standard matrix;

[0030] Query the maximum and minimum values ​​of each indicator in the standard matrix after deleting samples, and calculate the distance between each sample and the maximum value and the distance between each sample and the minimum value based on the maximum and minimum values;

[0031] For any sample, the total score is calculated based on the distance between the sample and the maximum value and the distance between the sample and the minimum value.

[0032] As a further solution of the present invention: the process of the positive conversion is:

[0033] For cost-type values,

[0034] For benefit-based values,

[0035] The calculation process of the total score is:

[0036] In the above content, y ij is the data of the i-th sample in the matrix of the selected area before the positive transformation under the j-th index, r ij is the positive data of the i-th sample in the matrix of the selected area under the j-th index; is the maximum value of the jth index, is the minimum value of the jth index; is the distance between the ith sample and the maximum value, is the distance between the i-th sample and the minimum value; is the maximum value of the jth index, is the minimum value of the jth indicator.

[0037] The technical solution of the present invention also provides an inland river LNG terminal intelligent site selection terminal, the terminal comprising:

[0038] A feature extraction module is used to receive the indicator type uploaded by the management party, query the area corresponding to each indicator type, take photos of the area, extract features from the photos of the area, and obtain the convolution kernel of each indicator type; wherein the indicator type includes two parameters, the value of one parameter includes at least guidance and constraint, and the value of the other parameter includes at least qualitative and quantitative; the convolution kernel is a preset image feature;

[0039] An image acquisition module, used to acquire a first remote sensing image of the area to be selected, determine a refined area based on the first remote sensing image, and acquire a second remote sensing image in the refined area; wherein the accuracy of the second remote sensing image is greater than the accuracy of the first remote sensing image;

[0040] A content extraction module, for extracting a quantitative area and a qualitative area in the first remote sensing image containing the second remote sensing image according to the obtained convolution kernel; the quantitative area is an area corresponding to the quantitative index in the spliced ​​close-up image, and the qualitative area is an area corresponding to the qualitative index in the spliced ​​close-up image;

[0041] The data acquisition module is used to identify the quantitative area and obtain the actual value of the quantitative indicator; count the qualitative area, establish a scoring questionnaire, send it to the scoring end, and receive the score feedback from the scoring end;

[0042] The data evaluation module is used to count all the scores, obtain the candidate area matrix, determine the total score of each sample based on the TOPSI S method, and determine the final candidate area based on the total score.

[0043] As a further solution of the present invention: the image acquisition module includes:

[0044] A contour recognition unit is used to obtain a first remote sensing image of the area to be selected, perform contour recognition on the first remote sensing image, and obtain a region contour;

[0045] A region marking unit, used for marking the region outline as a refined region when the area of ​​the region outline is included in a preset area range;

[0046] The area replacement unit is used to obtain a second remote sensing image of the refined area to replace the corresponding area of ​​the first remote sensing image.

[0047] As a further solution of the present invention: the content extraction module includes:

[0048] A matching unit, used for reading the convolution kernels of various indicator types, traversing the first remote sensing image replaced by the convolution kernel, and calculating the matching degree in real time;

[0049] A recording unit, configured to record a matching position when the matching degree reaches a preset matching degree threshold;

[0050] The classification unit is used to query the indicator type corresponding to the convolution kernel and divide the matching position into quantitative area and qualitative area according to the indicator type.

[0051] As a further solution of the present invention: the data evaluation module includes:

[0052] A type arrangement unit, used to arrange various indicator types according to a preset indicator sequence;

[0053] A sample generating unit, used for reading the identified actual value when the indicator type is a quantitative indicator, and reading the score when the indicator type is a qualitative indicator, as a sample;

[0054] The sample arrangement unit arranges the samples in the order of the reference time to obtain a matrix of areas to be selected;

[0055] The matrix augmentation unit is used to splice the candidate area matrices of different candidate areas and determine the total score of each sample based on the TOPS IS method;

[0056] The query unit is used to query the candidate area corresponding to the sample with the largest total score as the final candidate area.

[0057] Compared with the prior art, the beneficial effect of the present invention is that the present invention entrusts the information collection link to the satellite, and after acquiring the remote sensing images collected by the satellite, the information is extracted according to the preset indicator type, and only qualitative indicators are extracted and handed over to the expert group for evaluation, which greatly simplifies the workload of the expert group and enables the expert group to work remotely. This makes it possible to further expand the expert group at the same cost, improve efficiency, and also improve evaluation accuracy, which is a virtuous circle. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention.

[0059] Figure 1 This is a flowchart of the intelligent site selection method for inland LNG terminals.

[0060] Figure 2 This is the first sub-process flowchart of the intelligent site selection method for inland LNG terminals.

[0061] Figure 3 This is the second sub-process flowchart of the intelligent site selection method for inland LNG terminals.

[0062] Figure 4 This is the third sub-process flowchart of the intelligent site selection method for inland LNG terminals.

[0063] Figure 5This is the structural diagram of the intelligent site selection terminal for inland LNG terminals. DETAILED DESCRIPTION

[0064] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0065] Figure 1 The flowchart of the intelligent site selection method for an inland LNG terminal is as follows. In an embodiment of the present invention, an intelligent site selection method for an inland LNG terminal includes:

[0066] Step S100: receiving the indicator type uploaded by the management party, querying the area corresponding to each indicator type, taking photos of the area, extracting features from the photos of the area, and obtaining convolution kernels of each indicator type; wherein the indicator type includes two parameters, the value of one parameter includes at least guidance and constraint, and the value of the other parameter includes at least qualitative and quantitative; the convolution kernel is a preset image feature;

[0067] The application scenario of the technical solution of the present invention is to select the most suitable address from several regions to build an LNG terminal; this requires evaluation of several regions, and the evaluation aspects are pre-set by the management, that is, the indicator types in the above content, each indicator type corresponds to a different actual area, and the drone collects photos of the actual areas corresponding to different indicator types, and uses existing feature extraction methods to extract image features from the photos as convolution kernels; the convolution kernels are stored with the indicator type as the index, so that each indicator type corresponds to multiple convolution kernels (image features).

[0068] Regarding the classification process of the above indicator types, the indicator type is mainly represented by two parameters. One parameter indicates whether the indicator type is quantitative or qualitative. The meanings of quantitative and qualitative are clear and have the same meaning as the nouns themselves. The other parameter indicates whether the indicator type is directional or binding. Binding indicators correspond to the relevant standards and specifications issued by the country and are standards that must be met. Guiding indicators have an evaluation standard and their specific values ​​have a certain selection space.

[0069] In the prior art, indicator types include:

[0070]

[0071]

[0072] Table 1 Index type table

[0073] Table 1 above is one of the indicator types uploaded by the administrator.

[0074] Step S200: obtaining a first remote sensing image of the area to be selected, determining a refined area based on the first remote sensing image, and obtaining a second remote sensing image in the refined area; wherein the accuracy of the second remote sensing image is greater than the accuracy of the first remote sensing image;

[0075] The areas to be selected are several possible areas for site selection. First, obtain remote sensing images with lower accuracy, roughly scan the remote sensing images, and select some contours that need to be refined, which are called refined areas. Then, obtain remote sensing images with higher accuracy in the refined areas for subsequent analysis and processing.

[0076] The significance of the above-mentioned gradient processing process is that it is easier to obtain remote sensing images with lower precision, and they can be obtained with the help of existing map services. However, remote sensing images with higher precision often require the subject executing this method to have higher permissions to obtain them, and the acquisition cost is higher. In addition, in the subsequent processing process, only the more important areas (refined areas) have higher precision and more complicated processing, while other areas are relatively simple and the processing efficiency is very high.

[0077] Step S300: extracting a quantitative area and a qualitative area in the first remote sensing image containing the second remote sensing image according to the obtained convolution kernel; the quantitative area is an area corresponding to the quantitative index in the spliced ​​close-up image, and the qualitative area is an area corresponding to the qualitative index in the spliced ​​close-up image;

[0078] The convolution kernel obtained in step S100 is traversed in the first remote sensing image containing the second remote sensing image to match the regions corresponding to each indicator type. After the regions are matched, the regions are divided into quantitative regions and qualitative regions according to whether the corresponding indicator type is quantitative or qualitative.

[0079] Step S400: Identify the quantitative area and obtain the actual value of the quantitative indicator; count the qualitative area, establish a scoring questionnaire, send it to the scoring end, and receive the score feedback from the scoring end;

[0080] Since there is a definite standard for the quantitative area, when evaluating the quantitative area, the existing image recognition technology (mainly based on contour recognition algorithms) can be used to obtain the required dimensions in the quantitative area, and compare them with the preset standard values ​​to determine a score; for the qualitative area, since there is no definite standard, the Delphi method is used to score the guiding qualitative indicators of each alternative plan (3 rounds), and the scoring results are processed to convert the qualitative judgment into quantitative data; the Delphi method is to forward the data to the expert group for evaluation. From the perspective of this method, it is a simple data sending and receiving process, and the score fed back by the evaluation end can be received.

[0081] Step S500: Count all the scores to obtain a candidate area matrix, determine the total score of each sample based on the TOPSIS method, and determine the final candidate area based on the total score;

[0082] After the processing of step S400, all indicators will have a score, which constitutes a sample; for the same candidate area, close-up images at multiple times will be collected, so multiple samples can be constructed; the samples corresponding to each candidate area are counted to construct a candidate area matrix. The sample size is very large. Based on the TOPSI S method, the candidate area matrix is ​​evaluated to obtain the total score of each sample. By comparing the total scores of multiple samples, a final candidate area can be selected.

[0083] Figure 2 This is a flowchart of the first sub-process of the intelligent site selection method for an inland LNG terminal. The steps of acquiring a remote sensing image of the area to be selected, determining a collection path based on the remote sensing image, and acquiring a close-up image along the collection path by a drone include:

[0084] Step S201: acquiring a first remote sensing image of the area to be selected, and performing contour recognition on the first remote sensing image to obtain a region contour;

[0085] Step S202: when the area of ​​the region outline is included in the preset area range, marking the region outline as a refined region;

[0086] Step S203: Acquire a second remote sensing image of the refined area to replace the corresponding area of ​​the first remote sensing image.

[0087] In an example of the technical solution of the present invention, the acquisition process of the second remote sensing image is specifically limited. First, the first remote sensing image is acquired, contour recognition is performed in the first remote sensing image, and the regional contour is acquired. In the technical solution of the present invention, the contour of the area corresponding to the indicator is almost certain, and its area is roughly within a range. For example, the large area of ​​vegetation in the remote sensing image has little effect on the site selection of the wharf. It is invalid information and is eliminated through the contour area, and higher-precision information collection is not performed.

[0088] In layman's terms, this application selects an area within a certain range as a refined area.

[0089] Figure 3 The second sub-process flowchart of the method for intelligent site selection of an inland LNG terminal is as follows: the step of stitching the close-up image based on the acquisition path and extracting the quantitative area and the qualitative area from the stitched close-up image according to the obtained convolution kernel comprises:

[0090] Step S301: reading the convolution kernel of each indicator type, traversing the first remote sensing image after replacement by the convolution kernel, and calculating the matching degree in real time;

[0091] Step S302: When the matching degree reaches a preset matching degree threshold, the matching position is recorded;

[0092] Step S303: query the indicator type corresponding to the convolution kernel, and divide the matching position into a quantitative area and a qualitative area according to the indicator type.

[0093] The determined convolution kernel is read, and the convolution kernel traverses the first remote sensing image containing the second remote sensing image to match the area corresponding to each indicator type, which is called the matching position. It is determined whether the indicator type is quantitative or qualitative, so as to classify the regional position and obtain the quantitative area and the qualitative area.

[0094] Figure 4 This is a flowchart of the third sub-process of the intelligent site selection method for inland LNG terminals. The steps of counting all scores to obtain a matrix of candidate areas, determining the total score of each sample based on the TOPSIS method, and determining the final candidate area according to the total score include:

[0095] Step S501: Arrange each indicator type according to a preset indicator sequence;

[0096] Step S502: when the indicator type is a quantitative indicator, read the identified actual value; when the indicator type is a qualitative indicator, read the score as a sample;

[0097] Step S503: Arrange the samples in the order of the reference time to obtain a matrix of areas to be selected;

[0098] Step S504: splicing the candidate area matrices of different candidate areas, and determining the total score of each sample based on the TOPSI S method;

[0099] Step S505: query the to-be-selected area corresponding to the sample with the largest total score as the final to-be-selected area.

[0100] When setting the indicator types, the management party sets a default order and arranges the indicator types according to the indicator order (such as Table 1).

[0101] When the indicator type is a quantitative indicator, the actual value is read, and the actual value is obtained by identification and judgment of the existing image recognition algorithm. When the indicator type is a qualitative indicator, the score is read, and the score is obtained by the Delphi method and obtained by feedback from the expert group.

[0102] The samples are arranged in the order of the benchmark time to obtain the matrix of candidate areas. The matrix of candidate areas is analyzed based on the TOPSIS method to obtain the total score of each sample. The candidate area corresponding to the sample with the highest total score is queried as the final site selection result.

[0103] The step of splicing the candidate area matrices of different candidate areas and determining the total score of each sample based on the TOPSIS method includes:

[0104] All data are normalized to obtain a standard matrix;

[0105] Extract the constraint index in the standard matrix and obtain the constraint sub-matrix;

[0106] When any value in the constraint sub-matrix exceeds the preset condition, the sample corresponding to the value is deleted from the standard matrix;

[0107] Query the maximum and minimum values ​​of each indicator in the standard matrix after deleting samples, and calculate the distance between each sample and the maximum value and the distance between each sample and the minimum value based on the maximum and minimum values;

[0108] For any sample, the total score is calculated based on the distance between the sample and the maximum value and the distance between the sample and the minimum value.

[0109] Before analysis, the values ​​of all samples in the matrix of the selected area need to be positively converted. For qualitative indicators, the corresponding values ​​are scores, which are benefit indicators by default during expert group evaluation. For quantitative indicators, the corresponding values ​​are the actual values ​​identified, which may be cost indicators or benefit indicators. Therefore, they need to be positively converted. Benefit indicators are indicators whose values ​​are as large as possible, and cost indicators are indicators whose values ​​are as small as possible.

[0110] Specifically, the process of the positive transformation is:

[0111] For cost-type values,

[0112] For benefit-based values,

[0113] The calculation process of the total score is:

[0114] In the above content, y ij is the data of the i-th sample in the matrix of the selected area before the positive transformation under the j-th index, r ij is the positive data of the i-th sample in the matrix of the selected area under the j-th index; is the maximum value of the jth index, is the minimum value of the jth index; is the distance between the ith sample and the maximum value, is the distance between the i-th sample and the minimum value; is the maximum value of the jth index, is the minimum value of the jth indicator.

[0115] Figure 5 The figure is a composition structure diagram of an inland river LNG terminal intelligent site selection terminal. In an embodiment of the present invention, an inland river LNG terminal intelligent site selection terminal, the terminal 10 includes:

[0116] The feature extraction module 11 is used to receive the indicator type uploaded by the management party, query the area corresponding to each indicator type, take photos of the area, extract features from the photos of the area, and obtain the convolution kernel of each indicator type; wherein the indicator type includes two parameters, the value of one parameter includes at least guidance and constraint, and the value of the other parameter includes at least qualitative and quantitative; the convolution kernel is a preset image feature;

[0117] The image acquisition module 12 is used to acquire a first remote sensing image of the area to be selected, determine a refined area based on the first remote sensing image, and acquire a second remote sensing image in the refined area; wherein the accuracy of the second remote sensing image is greater than the accuracy of the first remote sensing image;

[0118] The content extraction module 13 is used to extract a quantitative area and a qualitative area in the first remote sensing image containing the second remote sensing image according to the obtained convolution kernel; the quantitative area is the area corresponding to the quantitative index in the spliced ​​close-up image, and the qualitative area is the area corresponding to the qualitative index in the spliced ​​close-up image;

[0119] The data acquisition module 14 is used to identify the quantitative area and obtain the actual value of the quantitative indicator; count the qualitative area, establish a scoring questionnaire, send it to the scoring end, and receive the score feedback from the scoring end;

[0120] The data evaluation module 15 is used to count all the scores to obtain the candidate area matrix, determine the total score of each sample based on the TOPS IS method, and determine the final candidate area according to the total score.

[0121] Furthermore, the image acquisition module 12 includes:

[0122] A contour recognition unit is used to obtain a first remote sensing image of the area to be selected, perform contour recognition on the first remote sensing image, and obtain a region contour;

[0123] A region marking unit, used for marking the region outline as a refined region when the area of ​​the region outline is included in a preset area range;

[0124] The area replacement unit is used to obtain a second remote sensing image of the refined area to replace the corresponding area of ​​the first remote sensing image.

[0125] Specifically, the content extraction module 13 includes:

[0126] A matching unit, used for reading the convolution kernels of various indicator types, traversing the first remote sensing image replaced by the convolution kernel, and calculating the matching degree in real time;

[0127] A recording unit, configured to record a matching position when the matching degree reaches a preset matching degree threshold;

[0128] The classification unit is used to query the indicator type corresponding to the convolution kernel and divide the matching position into quantitative area and qualitative area according to the indicator type.

[0129] In addition, the data evaluation module 15 comprises:

[0130] A type arrangement unit, used to arrange various indicator types according to a preset indicator sequence;

[0131] A sample generating unit, used for reading the identified actual value when the indicator type is a quantitative indicator, and reading the score when the indicator type is a qualitative indicator, as a sample;

[0132] The sample arrangement unit arranges the samples in the order of the reference time to obtain a matrix of areas to be selected;

[0133] The matrix augmentation unit is used to splice the candidate area matrices of different candidate areas and determine the total score of each sample based on the TOPS IS method;

[0134] The query unit is used to query the candidate area corresponding to the sample with the largest total score as the final candidate area.

[0135] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An intelligent site selection method for an inland LNG terminal, characterized in that: The method comprises: Receive the indicator type uploaded by the management party, query the area corresponding to each indicator type, take photos of the area, extract features from the photos of the area, and obtain the convolution kernel of each indicator type; wherein the indicator type includes two parameters, the value of one parameter includes at least guidance and constraint, and the value of the other parameter includes at least qualitative and quantitative; the convolution kernel is a preset image feature; Acquire a first remote sensing image of the area to be selected, determine a refined area based on the first remote sensing image, and acquire a second remote sensing image in the refined area; wherein the accuracy of the second remote sensing image is greater than the accuracy of the first remote sensing image; Extracting a quantitative area and a qualitative area in the first remote sensing image containing the second remote sensing image according to the obtained convolution kernel; the quantitative area is an area corresponding to the quantitative index in the spliced ​​close-up image, and the qualitative area is an area corresponding to the qualitative index in the spliced ​​close-up image; Identify the quantitative area and obtain the actual value of the quantitative indicator; count the qualitative area, establish a scoring questionnaire, send it to the scoring end, and receive the scoring feedback from the scoring end; All scores are counted to obtain the candidate area matrix, and the total score of each sample is determined based on the TOPSIS method, and the final candidate area is determined based on the total score; The steps of counting all the scores to obtain a matrix of candidate areas, determining the total score of each sample based on the TOPSIS method, and determining the final candidate areas according to the total score include: Arrange each indicator type according to the preset indicator order; When the indicator type is a quantitative indicator, read the identified actual value; when the indicator type is a qualitative indicator, read the score as a sample; Arrange the samples in the order of the reference time to obtain the matrix of the areas to be selected; The candidate area matrices of different candidate areas are spliced ​​together, and the total score of each sample is determined based on the TOPSIS method; Query the candidate area corresponding to the sample with the largest total score as the final candidate area; The step of splicing the candidate area matrices of different candidate areas and determining the total score of each sample based on the TOPSIS method includes: All data are normalized to obtain a standard matrix; Extract the constraint index in the standard matrix and obtain the constraint sub-matrix; When any value in the constraint sub-matrix exceeds the preset condition, the sample corresponding to the value is deleted from the standard matrix; Query the maximum and minimum values ​​of each indicator in the standard matrix after deleting samples, and calculate the distance between each sample and the maximum value and the distance between each sample and the minimum value based on the maximum and minimum values; For any sample, the total score is calculated based on the distance between the sample and the maximum value and the distance between the sample and the minimum value.

2. The intelligent site selection method for inland LNG terminal according to claim 1 is characterized in that: The steps of acquiring a first remote sensing image of the area to be selected, determining a refined area based on the first remote sensing image, and acquiring a second remote sensing image in the refined area include: Acquire a first remote sensing image of the area to be selected, perform contour recognition on the first remote sensing image, and obtain a regional contour; When the area of ​​the region outline is included in the preset area range, the region outline is marked as a refined area; A second remote sensing image of the refined area is obtained to replace the corresponding area of ​​the first remote sensing image.

3. The intelligent site selection method for inland LNG terminal according to claim 2 is characterized in that: The step of extracting the quantitative area and the qualitative area in the first remote sensing image containing the second remote sensing image according to the obtained convolution kernel comprises: Read the convolution kernel of each indicator type, traverse the first remote sensing image replaced by the convolution kernel, and calculate the matching degree in real time; When the matching degree reaches a preset matching degree threshold, recording the matching position; Query the indicator type corresponding to the convolution kernel, and divide the matching position into quantitative area and qualitative area according to the indicator type.

4. The intelligent site selection method for inland LNG terminal according to claim 1 is characterized in that: The process of the positive transformation is: For cost-type values, ; For benefit-based values, ; The calculation process of the total score is: ; ; ; in, is the data of the i-th sample in the matrix of the selected area before the positive transformation under the j-th index, is the positive data of the i-th sample in the matrix of the selected area under the j-th index; is the maximum value of the data before the positive transformation of the jth index, is the minimum value of the data before the positive transformation of the jth indicator; is the distance between the ith sample and the maximum value, is the distance between the i-th sample and the minimum value; is the maximum value of the positive data of the jth indicator, is the minimum value of the positive data of the jth indicator.

5. An intelligent site selection terminal for inland LNG terminal, characterized in that: The terminal comprises: A feature extraction module is used to receive the indicator type uploaded by the management party, query the area corresponding to each indicator type, take photos of the area, extract features from the photos of the area, and obtain the convolution kernel of each indicator type; wherein the indicator type includes two parameters, the value of one parameter includes at least guidance and constraint, and the value of the other parameter includes at least qualitative and quantitative; the convolution kernel is a preset image feature; An image acquisition module, used to acquire a first remote sensing image of the area to be selected, determine a refined area based on the first remote sensing image, and acquire a second remote sensing image in the refined area; wherein the accuracy of the second remote sensing image is greater than the accuracy of the first remote sensing image; A content extraction module, for extracting a quantitative area and a qualitative area in the first remote sensing image containing the second remote sensing image according to the obtained convolution kernel; the quantitative area is an area corresponding to the quantitative index in the spliced ​​close-up image, and the qualitative area is an area corresponding to the qualitative index in the spliced ​​close-up image; The data acquisition module is used to identify the quantitative area and obtain the actual value of the quantitative indicator; count the qualitative area, establish a scoring questionnaire, send it to the scoring end, and receive the score feedback from the scoring end; The data evaluation module is used to count all the scores, obtain the candidate area matrix, determine the total score of each sample based on the TOPSIS method, and determine the final candidate area based on the total score; The data evaluation module includes: A type arrangement unit, used to arrange various indicator types according to a preset indicator sequence; A sample generating unit, used for reading the identified actual value when the indicator type is a quantitative indicator, and reading the score when the indicator type is a qualitative indicator, as a sample; The sample arrangement unit arranges the samples in the order of the reference time to obtain a matrix of areas to be selected; The matrix augmentation unit is used to splice the candidate area matrices of different candidate areas and determine the total score of each sample based on the TOPSIS method; A query unit, used to query the candidate area corresponding to the sample with the largest total score as the final candidate area; The content of splicing the candidate area matrix of different candidate areas and determining the total score of each sample based on the TOPSIS method includes: All data are normalized to obtain a standard matrix; Extract the constraint index in the standard matrix and obtain the constraint sub-matrix; When any value in the constraint sub-matrix exceeds the preset condition, the sample corresponding to the value is deleted from the standard matrix; Query the maximum and minimum values ​​of each indicator in the standard matrix after deleting samples, and calculate the distance between each sample and the maximum value and the distance between each sample and the minimum value based on the maximum and minimum values; For any sample, the total score is calculated based on the distance between the sample and the maximum value and the distance between the sample and the minimum value.

6. The inland river LNG terminal intelligent site selection terminal according to claim 5 is characterized in that: The image acquisition module comprises: A contour recognition unit is used to obtain a first remote sensing image of the area to be selected, perform contour recognition on the first remote sensing image, and obtain a region contour; A region marking unit, used for marking the region outline as a refined region when the area of ​​the region outline is included in a preset area range; The area replacement unit is used to obtain a second remote sensing image of the refined area to replace the corresponding area of ​​the first remote sensing image.

7. The inland river LNG terminal intelligent site selection terminal according to claim 5, characterized in that: The content extraction module comprises: A matching unit, used for reading the convolution kernels of various index types, traversing the first remote sensing image replaced by the convolution kernel, and calculating the matching degree in real time; A recording unit, configured to record a matching position when the matching degree reaches a preset matching degree threshold; The classification unit is used to query the indicator type corresponding to the convolution kernel and divide the matching position into quantitative area and qualitative area according to the indicator type.

Images