diff --git a/adsingestp/parsers/elsevier.py b/adsingestp/parsers/elsevier.py
index 61a4d18..e68c51e 100644
--- a/adsingestp/parsers/elsevier.py
+++ b/adsingestp/parsers/elsevier.py
@@ -183,23 +183,19 @@ def _parse_title_abstract(self):
for abs in abs_all:
if abs.get("class", None) == "author":
abs_text_all = abs.find_all("ce:simple-para")
+ break
elif abs.find("ce:section-title"):
if abs.find("ce:section-title").get_text().lower() == "abstract":
abs_text_all = abs.find_all("ce:simple-para")
+ break
elif abs.find("ce:section-title").get_text().lower() == "highlights":
abs_text_all = abs.find_all("p")
- abstract = ""
- for abs_text in abs_text_all:
- abstract = (
- abstract
- + " "
- + self._detag(abs_text, self.HTML_TAGSET["abstract"]).strip()
- )
-
- if abstract:
- self.base_metadata["abstract"] = abstract
- break
+ abstract = ""
+ for abs_text in abs_text_all:
+ abstract = (
+ abstract + " " + self._detag(abs_text, self.HTML_TAGSET["abstract"]).strip()
+ )
if abstract:
self.base_metadata["abstract"] = self._clean_output(abstract)
diff --git a/tests/stubdata/input/els_abstract_author_1.xml b/tests/stubdata/input/els_abstract_author_1.xml
new file mode 100644
index 0000000..e77b82c
--- /dev/null
+++ b/tests/stubdata/input/els_abstract_author_1.xml
@@ -0,0 +1,1090 @@
+
+
+
+ application/xml
+ Merging experiment data and simulation data for parameter identification of shaft seal
+ Yuan Yin
+ Qiang He
+ Fengming Hu
+ Weifeng Huang
+
+
+ Acoustic emissions
+ Parameter identification
+ Transfer learning
+ Convolutional neural network
+
+
+ Measurement 236 (2024). doi:10.1016/j.measurement.2024.114863
+ journal
+ Measurement
+ © 2024 Published by Elsevier Ltd.
+ Elsevier Ltd
+ 0263-2241
+ 236
+ 15 August 2024
+ 2024-08-15
+ 10.1016/j.measurement.2024.114863
+ http://dx.doi.org/10.1016/j.measurement.2024.114863
+ doi:10.1016/j.measurement.2024.114863
+ 114863
+
+ 2024-05-09T00:00:00.000Z
+
+
+
+
+
+
+ Journals
+ S250.1
+
+
+
+ MEASUR
+ 114863
+ 114863
+ S0263-2241(24)00748-6
+ 10.1016/j.measurement.2024.114863
+
+
+
+
+ Fig. 1
+
+ Faults and the face contact.
+
+
+
+
+ Fig. 2
+
+ Overview of the acoustic emission signals acquired, taking experiment 2 as an example.
+
+
+
+
+ Fig. 3
+
+ Embedding frequency spectrums into low-dimensionality space with an auto-encoder.
+
+
+
+
+ Fig. 4
+
+ Distribution of the undetermined parameters.
+
+
+
+
+ Fig. 5
+
+ Some of the calculated results. (a)–(d) pressure fields at 4 different moments, taking case 1 as example; (e)
+
+
+
+ V
+
+
+ 0
+
+
+ output, taking 5 cases as example.
+
+
+
+
+
+ Fig. 6
+
+ Modified convolutional neural network architecture. The sequences are extended before filtered by convolution layers.
+
+
+
+
+ Fig. 7
+
+ Distribution of the identified load with filter size
+
+
+ s
+ =
+ 4
+
+ and embedding dimensionality
+
+
+ u
+ =
+ 64
+
+ . Medians (the points) and quartiles (the bars) are plotted.
+
+
+
+
+
+ Fig. 8
+
+ Comparison of transferring Module 1 from models using augmented data or not. (a) s=3, (b) s=4, (c) s=5.
+
+
+
+
+ Fig. 9
+
+ Comparison of different transfer modes when augmentation is applied. (a) s=3, (b) s=4, (c) s=5.
+
+
+
+
+ Fig. 10
+
+ Model output fluctuation along operation phases.
+
+
+
+
+ Table 1
+
+ Adjusted abnormities in the test rig.
+
+
+
+
+
+
+
+
+ Number of experiment
+ Load (N)
+ Rotor tilt (
+
+ μ
+ rad)
+
+ Group
+
+
+
+
+ 1
+ 29.7
+ 46
+ A
+
+
+ 2
+ 54.6
+ 46
+ A
+
+
+ 3
+ 79.5
+ 46
+ A
+
+
+ 4
+ 104.4
+ 46
+ A
+
+
+ 5
+ 129.3
+ 46
+ A
+
+
+ 6
+ 29.7
+ 144
+ B
+
+
+ 7
+ 54.6
+ 144
+ B
+
+
+ 8
+ 79.5
+ 144
+ B
+
+
+ 9
+ 104.4
+ 144
+ B
+
+
+ 10
+ 129.3
+ 144
+ B
+
+
+
+
+
+ Table 2
+
+ Varying parameters and their respective distributions for simulation data.
+
+
+
+
+
+
+
+ Dimension index
+
+ j
+
+
+ Variable
+ Distribution
+
+
+
+
+ 1
+ Extra axial force
+
+
+ F
+
+ (
+ N
+ )
+
+
+
+
+
+
+
+ N
+
+ (
+ 0
+ ,
+ 10
+
+
+ 0
+
+
+ 2
+
+
+ )
+
+
+
+
+ a
+
+
+
+
+ 2–3
+ Extra moment
+
+
+
+
+ M
+
+
+ x
+
+
+ ,
+
+
+ M
+
+
+ y
+
+
+
+
+ (
+ N
+
+ m
+ )
+
+
+
+
+
+
+
+ N
+
+ (
+ 0
+ ,
+
+
+ 5
+
+
+ 2
+
+
+ )
+
+
+
+
+
+
+ 4
+ Support stiffness and damping scale
+
+
+
+ Q
+
+
+ s
+
+
+
+
+
+
+
+
+
+ N
+
+
+ +
+
+
+
+ (
+ 0
+ ,
+ 2
+
+
+ 5
+
+
+ 2
+
+
+ )
+
+
+
+
+ b
+
+
+
+
+ 5–6
+ Rotor tilt
+
+
+
+
+ γ
+
+
+ r
+ x
+ 0
+
+
+ ,
+
+
+ γ
+
+
+ r
+
+
+ y
+
+
+ 0
+
+
+
+
+
+
+ (
+ mrad
+ )
+
+
+
+
+
+
+
+ N
+
+ (
+ 0
+ ,
+ 0
+ .
+ 2
+
+
+ 5
+
+
+ 2
+
+
+ )
+
+
+
+
+
+
+ 7–14
+ Waviness
+
+
+
+
+ h
+
+
+ ws
+ 2
+
+
+ ,
+ …
+ ,
+
+
+ h
+
+
+ wr
+ 3
+
+
+
+
+ (
+ μ
+ m
+ )
+
+
+
+
+
+
+
+ N
+
+ (
+ 0
+ ,
+ 0
+ .
+ 2
+
+
+ 5
+
+
+ 2
+
+
+ )
+
+
+
+
+
+
+ 15
+ Coning
+
+
+ δ
+
+
+ (
+ μ
+ m
+ )
+
+
+
+
+
+
+
+ N
+
+ (
+ 0
+ ,
+ 0
+ .
+ 2
+
+
+ 5
+
+
+ 2
+
+
+ )
+
+
+
+
+
+
+ 16
+ Speed
+
+
+ ω
+
+
+ (
+ rad
+ )
+
+
+
+
+
+
+
+ U
+
+ (
+ 600
+ ,
+ 1200
+ )
+
+
+
+
+ c
+
+
+
+
+
+
+ a
+
+
+
+ N
+
+ (
+ μ
+ ,
+
+
+ σ
+
+
+ 2
+
+
+ )
+
+
+ denotes a normal distribution with mean
+
+ μ
+ and standard variance
+
+ σ
+ .
+
+
+
+ b
+
+
+
+
+
+ N
+
+
+ +
+
+
+
+ (
+ μ
+ ,
+
+
+ σ
+
+
+ 2
+
+
+ )
+
+
+ denotes the right half (in terms of probability density function) of a normal distribution with mean
+
+ μ
+ and standard variance
+
+ σ
+ .
+
+
+
+ c
+
+
+
+ U
+
+ (
+ a
+ ,
+ b
+ )
+
+
+ denotes a uniform distribution between
+
+ a
+ and
+
+ b
+ .
+
+
+
+
+ Table 3
+
+ Cross validation error of different models.
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+
+ s
+
+
+
+
+ u
+
+
+ Not augmented
+ Augmented
+
+
+
+
+ –
+ M1
+ M2
+ M1, M2
+ –
+ M1
+ M2
+ M1, M2
+
+
+
+
+ 3
+ 16
+ 0.291
+ 0.321
+ 0.29
+ 0.192
+ 0.291
+ 0.193
+ 0.351
+ 0.544
+
+
+
+ 32
+ 0.297
+ 0.255
+ 0.28
+ 0.333
+ 0.297
+ 0.225
+ 0.297
+ 0.244
+
+
+
+ 64
+ 0.259
+ 0.254
+ 0.344
+ 0.151
+ 0.259
+ 0.162
+ 0.137
+ 0.323
+
+
+
+ 128
+ 0.213
+ 0.246
+ 0.236
+ 0.218
+ 0.213
+ 0.158
+ 0.267
+ 0.243
+
+
+
+ 256
+ 0.194
+ 0.227
+ 0.227
+ 0.264
+ 0.194
+ 0.147
+ 0.203
+ 0.321
+
+
+ 4
+ 16
+ 0.279
+ 0.278
+ 0.378
+ 0.207
+ 0.279
+ 0.272
+ 0.198
+ 0.298
+
+
+
+ 32
+ 0.262
+ 0.258
+ 0.311
+ 0.526
+ 0.262
+ 0.175
+ 0.317
+ 0.201
+
+
+
+ 64
+ 0.244
+ 0.257
+ 0.265
+ 0.229
+ 0.244
+ 0.157
+ 0.211
+ 0.221
+
+
+
+ 128
+ 0.270
+ 0.221
+ 0.267
+ 0.246
+ 0.270
+ 0.130
+ 0.267
+ 0.215
+
+
+
+ 256
+ 0.220
+ 0.262
+ 0.204
+ 0.172
+ 0.220
+ 0.133
+ 0.258
+ 0.254
+
+
+ 5
+ 16
+ 0.311
+ 0.324
+ 0.512
+ 0.782
+ 0.311
+ 0.236
+ 0.292
+ 0.721
+
+
+
+ 32
+ 0.305
+ 0.265
+ 0.302
+ 0.496
+ 0.305
+ 0.175
+ 0.294
+ 0.534
+
+
+
+ 64
+ 0.275
+ 0.257
+ 0.275
+ 0.481
+ 0.275
+ 0.169
+ 0.270
+ 0.392
+
+
+
+ 128
+ 0.211
+ 0.252
+ 0.287
+ 0.304
+ 0.211
+ 0.164
+ 0.258
+ 0.237
+
+
+
+ 256
+ 0.208
+ 0.243
+ 0.219
+ 0.210
+ 0.208
+ 0.135
+ 0.281
+ 0.314
+
+
+
+
+
+
+ Merging experiment data and simulation data for parameter identification of shaft seal
+
+
+ Yuan
+ Yin
+ Writing – original draft
+ Visualization
+ Methodology
+ Investigation
+ Formal analysis
+
+
+ Qiang
+ He
+ Visualization
+ Software
+ Conceptualization
+
+
+ Fengming
+ Hu
+ Writing – review & editing
+ Investigation
+ Data curation
+
+
+ Weifeng
+ Huang
+ Writing – review & editing
+ Supervision
+ Resources
+ Project administration
+ Investigation
+ Funding acquisition
+ Conceptualization
+
+ ⁎
+
+ huangwf@tsinghua.edu.cn
+
+
+ State Key Laboratory of Tribology in Advanced Equipment, Tsinghua University, Beijing 100084, China
+
+ State Key Laboratory of Tribology in Advanced Equipment, Tsinghua University
+ Beijing
+ 100084
+ China
+
+ State Key Laboratory of Tribology in Advanced Equipment, Tsinghua University, Beijing 100084, China
+
+
+ ⁎
+ Corresponding author.
+
+
+
+
+
+
+ Abstract
+
+ A parameter identification method which merges experimentally monitored signals and physics-based simulation data is proposed, targeting the identification tasks in shaft seals which are challenging due to the high-dimensional parameter space. Features were extracted from the monitored acoustic emission signals following a proposed cross-timescale analysis routine, and the simulation data were augmented using Kriging surrogate model to obtain a dataset with stratified fidelity. Then, a transferable architecture of convolutional neural network modified for periodical data was proposed, with which part of the parameters trained by simulation data were reserved when training model using acoustic emission data acquired in experiments. Cross validation shows that transfer learning can effectively improve the performance, provided data augmentation and proper transfer mode. In conclusion, the study provides an effective parameter identification method which merges the simulation data which carry the physical knowledge and the experiment data which carry directly monitored results.
+
+
+
+ Highlights
+
+
+
+
+ •
+ A method of performing parameter identification of shaft seal with acoustic emission monitoring is proposed. The method uses low-cost simulation data to pre-train, and uses reliable experiment data to fine-tune.
+
+
+ •
+ The manuscript describes the concrete measures for a good transferring, including introducing surrogate-model-based augmentation, performing auto-encoder to embed frequency spectrums, and designing the neural network architecture. These could be enlightening for audiences who are working on similar tasks.
+
+
+
+
+
+
+ Keywords
+
+ Acoustic emissions
+
+
+ Parameter identification
+
+
+ Transfer learning
+
+
+ Convolutional neural network
+
+
+
+ Data availability
+ The data that has been used is confidential.
+
+
+
+
+ text body here
+
+
+ Acknowledgments
+ This work is supported by ABCDE.
+
+
+
+
+
+ References
+
+
+ [1]
+
+
+
+
+ Abc
+ D.
+
+
+
+ Reference paper
+
+
+
+
+
+
+ XYZ.
+
+ 13
+
+ 2021
+
+ 1234567890
+
+
+ D. Abc, Reference paper, XYZ 13 (2021) 1234567890.
+
+
+
+
+
+
diff --git a/tests/stubdata/output/els_abstract_author_1.json b/tests/stubdata/output/els_abstract_author_1.json
new file mode 100644
index 0000000..fce3d33
--- /dev/null
+++ b/tests/stubdata/output/els_abstract_author_1.json
@@ -0,0 +1,135 @@
+{
+ "abstract": {
+ "textEnglish": "A parameter identification method which merges experimentally monitored signals and physics-based simulation data is proposed, targeting the identification tasks in shaft seals which are challenging due to the high-dimensional parameter space. Features were extracted from the monitored acoustic emission signals following a proposed cross-timescale analysis routine, and the simulation data were augmented using Kriging surrogate model to obtain a dataset with stratified fidelity. Then, a transferable architecture of convolutional neural network modified for periodical data was proposed, with which part of the parameters trained by simulation data were reserved when training model using acoustic emission data acquired in experiments. Cross validation shows that transfer learning can effectively improve the performance, provided data augmentation and proper transfer mode. In conclusion, the study provides an effective parameter identification method which merges the simulation data which carry the physical knowledge and the experiment data which carry directly monitored results."
+ },
+ "authors": [
+ {
+ "affiliation": [
+ {
+ "affPubRaw": "State Key Laboratory of Tribology in Advanced Equipment, Tsinghua University, Beijing 100084, China"
+ }
+ ],
+ "name": {
+ "given_name": "Yuan",
+ "surname": "Yin"
+ }
+ },
+ {
+ "affiliation": [
+ {
+ "affPubRaw": "State Key Laboratory of Tribology in Advanced Equipment, Tsinghua University, Beijing 100084, China"
+ }
+ ],
+ "name": {
+ "given_name": "Qiang",
+ "surname": "He"
+ }
+ },
+ {
+ "affiliation": [
+ {
+ "affPubRaw": "State Key Laboratory of Tribology in Advanced Equipment, Tsinghua University, Beijing 100084, China"
+ }
+ ],
+ "name": {
+ "given_name": "Fengming",
+ "surname": "Hu"
+ }
+ },
+ {
+ "affiliation": [
+ {
+ "affPubRaw": "State Key Laboratory of Tribology in Advanced Equipment, Tsinghua University, Beijing 100084, China"
+ }
+ ],
+ "attrib": {
+ "email": "huangwf@tsinghua.edu.cn"
+ },
+ "name": {
+ "given_name": "Weifeng",
+ "surname": "Huang"
+ }
+ }
+ ],
+ "copyright": {
+ "statement": "\u00a9 2024 Published by Elsevier Ltd.",
+ "status": true
+ },
+ "doctype": "article",
+ "editorialHistory": {
+ "acceptedDate": "2024-05-06",
+ "receivedDates": [
+ "2024-01-31"
+ ],
+ "revisedDates": [
+ "2024-03-30"
+ ]
+ },
+ "esources": [
+ {
+ "location": "http://dx.doi.org/10.1016/j.measurement.2024.114863",
+ "source": "pub_html"
+ }
+ ],
+ "keywords": [
+ {
+ "keyString": "Acoustic emissions",
+ "keySystem": "Elsevier"
+ },
+ {
+ "keyString": "Parameter identification",
+ "keySystem": "Elsevier"
+ },
+ {
+ "keyString": "Transfer learning",
+ "keySystem": "Elsevier"
+ },
+ {
+ "keyString": "Convolutional neural network",
+ "keySystem": "Elsevier"
+ }
+ ],
+ "pagination": {
+ "electronicID": "114863"
+ },
+ "persistentIDs": [
+ {
+ "DOI": "10.1016/j.measurement.2024.114863"
+ }
+ ],
+ "pubDate": {
+ "printDate": "2024-08-15"
+ },
+ "publication": {
+ "ISSN": [
+ {
+ "issnString": "0263-2241",
+ "pubtype": "not specified"
+ }
+ ],
+ "pubName": "Measurement",
+ "pubYear": "2024",
+ "publisher": "Elsevier Ltd",
+ "volumeNum": "236"
+ },
+ "publisherIDs": [
+ {
+ "Identifier": "S0263-2241(24)00748-6",
+ "attribute": "PII"
+ }
+ ],
+ "recordData": {
+ "createdTime": "",
+ "loadFormat": "OtherXML",
+ "loadLocation": "",
+ "loadType": "fromFile",
+ "parsedTime": "",
+ "recordOrigin": ""
+ },
+ "references": [
+ " Abc D. Reference paper XYZ. 13 2021 1234567890 "
+ ],
+ "title": {
+ "textEnglish": "Merging experiment data and simulation data for parameter identification of shaft seal"
+ }
+}
diff --git a/tests/test_elsevier.py b/tests/test_elsevier.py
index 6c5d3df..03bfe2d 100644
--- a/tests/test_elsevier.py
+++ b/tests/test_elsevier.py
@@ -38,6 +38,7 @@ def test_elsevier(self):
"els_odd_cover_date",
"els_roman_num_1",
"els_roman_num_2",
+ "els_abstract_author_1",
]
for f in filenames:
test_infile = os.path.join(self.inputdir, f + ".xml")