CN111881950A - Method and device for representing characteristics of current time sequence of turnout switch machine - Google Patents
Method and device for representing characteristics of current time sequence of turnout switch machine Download PDFInfo
- Publication number
- CN111881950A CN111881950A CN202010664463.3A CN202010664463A CN111881950A CN 111881950 A CN111881950 A CN 111881950A CN 202010664463 A CN202010664463 A CN 202010664463A CN 111881950 A CN111881950 A CN 111881950A
- Authority
- CN
- China
- Prior art keywords
- symbol
- current time
- sequence
- time sequence
- current
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 44
- 238000013528 artificial neural network Methods 0.000 claims abstract description 54
- 238000000605 extraction Methods 0.000 claims description 23
- 239000013598 vector Substances 0.000 claims description 15
- 238000004590 computer program Methods 0.000 claims description 12
- 238000012549 training Methods 0.000 claims description 10
- 230000002123 temporal effect Effects 0.000 claims description 2
- 230000003068 static effect Effects 0.000 abstract description 10
- 230000002159 abnormal effect Effects 0.000 abstract description 9
- 238000001514 detection method Methods 0.000 abstract description 3
- 238000004891 communication Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 6
- 238000006243 chemical reaction Methods 0.000 description 5
- 238000012512 characterization method Methods 0.000 description 4
- 239000000284 extract Substances 0.000 description 4
- 238000010801 machine learning Methods 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000012163 sequencing technique Methods 0.000 description 2
- 208000027418 Wounds and injury Diseases 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 230000006378 damage Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 208000014674 injury Diseases 0.000 description 1
- 238000012423 maintenance Methods 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/21—Design or setup of recognition systems or techniques; Extraction of features in feature space; Blind source separation
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- Life Sciences & Earth Sciences (AREA)
- Artificial Intelligence (AREA)
- General Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Software Systems (AREA)
- Mathematical Physics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Computing Systems (AREA)
- Molecular Biology (AREA)
- General Health & Medical Sciences (AREA)
- Evolutionary Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Image Analysis (AREA)
Abstract
The embodiment of the invention provides a method and a device for representing characteristics of a current time sequence of a switch machine, wherein the method comprises the following steps: counting a current time sequence of a turnout switch machine to obtain statistical characteristics of the current time sequence; extracting the characteristics of the current time sequence based on a deep neural network to obtain the time sequence characteristics of the current time sequence; and combining the statistical characteristics and the time sequence characteristics of the current time sequence, and taking the combined result as the representation of the current time sequence. The current time sequence is characterized by comprehensively using the static characteristics and the dynamic characteristics, and the relevant characteristics of the current time sequence are better characterized, so that the abnormal detection of the turnout switch machine can be more accurately carried out, and the fault type of the turnout switch machine can be more accurately identified.
Description
Technical Field
The invention relates to the technical field of intelligent operation and maintenance of urban rail transit, in particular to a method and a device for representing characteristics of a current time sequence of a switch point.
Background
The switch machine is a switch device of the switch, and can be used for switching the switch or locking the switch. Failure of a switch machine may result in derailment of an on-track train, resulting in significant economic loss and loss of life and personal injury. The current time sequence of the turnout switch machine can be used for judging whether the turnout switch machine is in a normal working state. If the working state is abnormal, further judging which abnormal working state is in.
The core of judging the working state of the turnout switch machine according to the current time sequence of the turnout switch machine is a characterization method aiming at a current curve. The working state of the turnout switch machine can be diagnosed more accurately based on the good representation; based on the poor representation, the operating state of the switch machine may be misjudged. The traditional characterization method aiming at the time series data is often analyzed from the statistical point of view. For example, the maximum value, the minimum value or the average value of the time series data is extracted to form the representation of the time series data. The method models the characteristics of the time series data from the static point of view, and the curves formed by the time series data are likely to be greatly different under the condition that the extracted statistical characteristics are the same.
Therefore, the current time sequence of the turnout switch machine is represented inaccurately by using the traditional time sequence data representation method, and the difference between different current time sequences cannot be distinguished accurately, so that the working state of the turnout switch machine cannot be diagnosed accurately according to the current time sequence representation of the turnout switch machine.
Disclosure of Invention
The embodiment of the invention provides a method and a device for representing characteristics of a current time sequence of a turnout switch machine, which are used for solving the defects that the representation of the current time sequence of the turnout switch machine is not accurate and the difference between different current time sequences cannot be accurately distinguished in the prior art and realizing the accurate representation of the current time sequence of the turnout switch machine.
The embodiment of the invention provides a method for representing characteristics of a current time sequence of a switch machine, which comprises the following steps:
counting a current time sequence of a turnout switch machine to obtain statistical characteristics of the current time sequence;
extracting the characteristics of the current time sequence based on a deep neural network to obtain the time sequence characteristics of the current time sequence;
and combining the statistical characteristics and the time sequence characteristics of the current time sequence, and taking the combined result as the representation of the current time sequence.
According to the feature representation method of the current time sequence of the turnout switch machine, the feature extraction is carried out on the current time sequence based on the deep neural network, and the step of obtaining the time sequence feature of the current time sequence comprises the following steps:
converting the current time series into a symbol series; wherein the symbol sequence is a sequence represented by a plurality of symbols;
performing one-hot encoding on the symbol sequence;
and extracting the characteristic of the one-hot code of the symbol sequence based on the deep neural network to obtain the time sequence characteristic of the current time sequence.
According to one embodiment of the method for characterizing the current time series of the turnout switch machine, the step of converting the current time series into the symbol sequence comprises the following steps:
representing the current value of each moment in the current time sequence by a symbol; wherein the same current value is represented by the same symbol, and different current values are represented by different symbols.
According to one embodiment of the method for characterizing the current time series of the turnout switch machine, the step of converting the current time series into the symbol sequence comprises the following steps:
acquiring a current curve of the turnout switch machine according to the current time sequence;
dividing the current curve into a plurality of line segments, and calculating the current average value of each line segment;
representing each line segment by a symbol according to the current average value of each line segment; wherein the line segments with the same current average value are represented by the same symbol, and the line segments with different current average values are represented by different symbols.
According to one embodiment of the method for characterizing the current time series of the switch points, the step of encoding the symbol sequence by one hot includes:
for any symbol in the symbol sequence, setting the state of a state bit corresponding to the symbol in a preset state bit sequence as valid, and setting other state bits except the state bit corresponding to the symbol in the state bit sequence as invalid; wherein each state bit in the state bit sequence corresponds to a symbol.
According to one embodiment of the invention, the method for representing the characteristics of the current time series of the turnout switch machine performs characteristic extraction on the one-hot code of the symbol sequence based on the deep neural network, and the step of acquiring the time series characteristics of the current time series comprises the following steps:
for any symbol in the symbol sequence, taking the one-hot coding of the context of the symbol in the one-hot coding of the symbol sequence as the input of the deep neural network, and training the deep neural network so that the output layer of the deep neural network outputs the one-hot coding of the symbol in the one-hot coding of the symbol sequence;
taking the output of a hidden layer before the output layer in the trained deep neural network as the vector representation of the symbol;
and taking the average value of the vector representation of all the symbols as the time sequence characteristic of the current time sequence.
According to an embodiment of the present invention, the method for characterizing a time series of switch machine currents, for any symbol in the symbol sequence, using a one-hot coding of a context of the symbol in the one-hot coding of the symbol sequence as an input of the deep neural network, and before the step of training the deep neural network, further includes:
for any symbol in the symbol sequence, taking all other symbols except the symbol in a window which takes the symbol as the center in the symbol sequence as the context of the symbol; wherein the size of the window is preset.
The embodiment of the invention also provides a characteristic representation device of a current time sequence of a switch machine, which comprises the following components:
the statistical module is used for carrying out statistics on the current time sequence of the turnout switch machine to obtain the statistical characteristics of the current time sequence;
the extraction module is used for extracting the characteristics of the current time sequence based on a deep neural network to obtain the time sequence characteristics of the current time sequence;
and the merging module is used for merging the statistical characteristics and the time sequence characteristics of the current time sequence and taking the merging result as the representation of the current time sequence.
The embodiment of the present invention further provides an electronic device, which includes a memory, a processor and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the method for characterizing a time series of a switch machine current as described in any one of the above mentioned methods when executing the program.
Embodiments of the present invention further provide a non-transitory computer readable storage medium, on which a computer program is stored, the computer program, when executed by a processor, implementing the steps of the method for characterizing a temporal sequence of current of a switch machine of a switch point as described in any one of the above.
According to the method and the device for representing the characteristics of the current time sequence of the turnout switch machine, provided by the embodiment of the invention, the statistical characteristics are obtained by counting the current time sequence of the turnout switch machine from the global and static angles in the statistical characteristic extraction part; the time sequence feature extraction part extracts the time sequence feature of the current time sequence from a dynamic angle based on a deep neural network, and combines the two part features to be used as the feature representation of the current time sequence of the turnout switch machine, so that the static feature and the dynamic feature are comprehensively used for representing the current time sequence, the relevant characteristics of the current time sequence are better represented, and accordingly, the turnout switch machine can be more accurately detected in an abnormal mode and the fault type of the turnout switch machine can be more accurately identified.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and those skilled in the art can also obtain other drawings according to the drawings without creative efforts.
Fig. 1 is a schematic flow chart of a method for characterizing a current time sequence of a switch machine according to an embodiment of the present invention;
fig. 2 is a schematic diagram of a method for characterizing a current time sequence of a switch machine according to an embodiment of the present invention, in which the current time sequence is converted into a symbol sequence;
fig. 3 is a schematic diagram of a symbol sequence being subjected to one-hot encoding in a method for characterizing a switch machine current time sequence according to an embodiment of the present invention;
fig. 4 is a schematic diagram of feature extraction performed on a symbol sequence in a method for representing features of a switch machine current time sequence according to an embodiment of the present invention;
fig. 5 is a schematic structural diagram of a feature representation device of a switch machine current time sequence according to an embodiment of the present invention;
fig. 6 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
A method for characterizing a switch machine current time series according to an embodiment of the present invention is described below with reference to fig. 1, including: s101, counting a current time sequence of a turnout switch machine to obtain statistical characteristics of the current time sequence;
the current time sequence is a sequence formed by sequencing the current of the turnout switch machine at each moment in a certain historical time period according to the sequence of the acquisition time. And counting the current values in the current time sequence, and taking the characteristics obtained by counting as statistical characteristics. The statistical features include one or more of a current maximum, a current minimum, and a current average.
S102, extracting the characteristics of the current time sequence based on a deep neural network to obtain the time sequence characteristics of the current time sequence;
the present embodiment is not limited to the kind of deep neural network. A timing characteristic is a characteristic that reflects the timing characteristics of a current in a current time series. And extracting the time sequence characteristics of the current time sequence by using a deep neural network.
S103, combining the statistical characteristics and the time sequence characteristics of the current time sequence, and taking the combined result as the representation of the current time sequence.
The representation of the current time sequence obtained in the embodiment can be applied to judging whether the working state of the turnout switch machine is normal or not according to the representation of the current time sequence by using a machine learning model. And if the working state of the turnout switch machine is abnormal, further determining the fault type of the turnout switch machine by using a machine learning model.
The statistical and timing characteristics of the current time series may specifically be combined in a stitching manner. The statistical characteristics and the time sequence characteristics are combined and integrated, and the integrated characteristics are used for representing the current time sequence, so that the representation of the current time sequence not only comprises static statistical characteristics, but also comprises dynamic time sequence characteristics. The combined characteristics can better represent the relevant characteristics of the current time sequence, so that the abnormal detection can be more accurately carried out on the turnout switch machine and the fault type of the turnout switch machine can be identified.
In the embodiment, the statistical characteristics are obtained by counting the current time sequence of the turnout switch machine from the global and static angles in the statistical characteristic extraction part; the time sequence feature extraction part extracts the time sequence feature of the current time sequence from a dynamic angle based on a deep neural network, and combines the two part features to be used as the feature representation of the current time sequence of the turnout switch machine, so that the static feature and the dynamic feature are comprehensively used for representing the current time sequence, the relevant characteristics of the current time sequence are better represented, and accordingly, the turnout switch machine can be more accurately detected in an abnormal mode and the fault type of the turnout switch machine can be more accurately identified.
On the basis of the foregoing embodiment, in this embodiment, feature extraction is performed on the current time series based on a deep neural network, and the step of obtaining the time series feature of the current time series includes: converting the current time series into a symbol series; wherein the symbol sequence is a sequence represented by a plurality of symbols;
specifically, the current time series is first symbolized, that is, the current time series is converted into a series represented by a series of symbols. The definition of the symbol is set according to the actual situation.
Performing one-hot encoding on the symbol sequence;
the current time series is one-hot encoded after it is signed. One-hot coding, also known as one-bit-efficient coding, is a method of coding states using a given number of state bits, each state having independent state bits, and only one of which is active at any one time.
And extracting the characteristic of the one-hot code of the symbol sequence based on the deep neural network to obtain the time sequence characteristic of the current time sequence.
On the basis of the foregoing embodiment, the step of converting the current time series into a symbol series in this embodiment includes: representing the current value of each moment in the current time sequence by a symbol; wherein the same current value is represented by the same symbol, and different current values are represented by different symbols.
Specifically, the present embodiment represents the current value at each time in the current time series by one symbol, which is analogous to a word in natural language. The sign of the current value at each time may be represented by the current value at that time, or may be represented by another number or word. The present embodiment is not limited to the representation of symbols.
On the basis of the foregoing embodiment, the step of converting the current time series into a symbol series in this embodiment includes: acquiring a current curve of the turnout switch machine according to the current time sequence;
as shown in fig. 2, a current curve is plotted according to a current time series with a current as a vertical axis and time as a horizontal axis, and circles are abstract symbols.
Dividing the current curve into a plurality of line segments, and calculating the current average value of each line segment;
when the number of current values in the current time series is large, in order to reduce the calculation amount, the current curve is divided into a plurality of discrete intervals, and each discrete interval is represented by one symbol. Specifically, a time period from the earliest time to the latest time of obtaining the current in the current time sequence is divided into a plurality of sub-time periods. And counting the current average value of the line segment of the current curve in each sub-period. The current curve can be equally divided into a plurality of line segments, and the current average value of each line segment can be counted.
Representing each line segment by a symbol according to the current average value of each line segment; wherein the line segments with the same current average value are represented by the same symbol, and the line segments with different current average values are represented by different symbols.
The current average value of each line segment is represented by a symbol, and the symbol of the current average value of each line segment can be represented by the current value at the moment, and can also be represented by other numbers or words, so that the current time sequence is converted into a symbol sequence. The present embodiment is not limited to the representation of symbols. Suppose that the symbol sequence of the ith current time sequence isWhere each element in the sequence of symbols is a specific symbol.
On the basis of the foregoing embodiment, in this embodiment, the step of performing one-hot encoding on the symbol sequence includes: for any symbol in the symbol sequence, setting the state of a state bit corresponding to the symbol in a preset state bit sequence as valid, and setting other state bits except the state bit corresponding to the symbol in the state bit sequence as invalid; wherein each state bit in the state bit sequence corresponds to a symbol.
Specifically, as shown in fig. 3, the symbol sequence [5,2,3,2,1,4] is one-hot encoded. There are 5 status bits in the predetermined sequence of status bits. Wherein the first state bit corresponds to symbol 5, the second state bit corresponds to symbol 4, the third state bit corresponds to symbol 3, the fourth state bit corresponds to symbol 2, and the fifth state bit corresponds to symbol 1. For symbol 5, the first state bit corresponding to symbol 5 is set to 1, and the other state bits are set to 0. 1 indicates valid and 0 indicates invalid.
After one-hot coding, the symbol sequenceConversion into one-hot coded symbol sequencesWhereinK is the number of state bits in the state bit sequence.
On the basis of the foregoing embodiment, in this embodiment, feature extraction is performed on the one-hot code of the symbol sequence based on the deep neural network, and the step of obtaining the time sequence feature of the current time sequence includes: for any symbol in the symbol sequence, taking the one-hot coding of the context of the symbol in the one-hot coding of the symbol sequence as the input of the deep neural network, and training the deep neural network so that the output layer of the deep neural network outputs the one-hot coding of the symbol in the one-hot coding of the symbol sequence;
in particular, one-hot encoding of symbol sequencesCarrying out deep neural network by taking one-hot coding of context of jth symbol as training sampleAnd (5) training. The context of the jth symbol is a number of symbols in the symbol sequence that precede and/or follow the jth symbol and are immediately adjacent to the jth symbol. The one-hot encoding of the context of the jth symbol is taken as input,as an output of the deep neural network, the previous layer of the output layer is a hidden layer, and the deep neural network is constructed, as shown in fig. 4.
Taking the output of a hidden layer before the output layer in the trained deep neural network as the vector representation of the symbol; and taking the average value of the vector representation of all the symbols as the time sequence characteristic of the current time sequence.
The output of a previous hidden layer of the trained output layer of the deep neural networkAs a vector representation of the jth symbol. In this way a vector representation of each symbol is obtained. And taking the average value of vector representations of all symbols in the symbol sequence as the time sequence characteristic of the current time sequence. Average of vector representations of all symbols in a sequence of symbolsAll timing information in the symbol sequence is contained. Where M is the length of the vector representation.
Statistical characterization of the ith current time series of a switch machineWherein N is the number of statistical features. The timing characteristic is hiSplicing the two characteristics to obtainThus, the dynamic time sequence characteristic and the static statistical characteristic of the current time sequence of the turnout switch machine are codedIn this fixed length vector. The coded feature vectors can be used for training an anomaly detection algorithm and a fault mode identification algorithm, so that the working state and the fault category of the turnout switch machine can be quickly and accurately identified.
On the basis of the foregoing embodiment, in this embodiment, regarding any symbol in the symbol sequence, the one-hot coding of the context of the symbol in the one-hot coding of the symbol sequence is used as the input of the deep neural network, and the step of training the deep neural network further includes: for any symbol in the symbol sequence, taking all other symbols except the symbol in a window which takes the symbol as the center in the symbol sequence as the context of the symbol; wherein the size of the window is preset.
Specifically, given a window size of w, the context in which the jth symbol is obtained isAs shown in fig. 4, willAs input to the deep neural network.
The following describes the feature representation device of the current time series of the switch points provided by the embodiment of the present invention, and the feature representation device of the current time series of the switch points described below and the feature representation method of the current time series of the switch points described above may be referred to correspondingly.
As shown in fig. 5, the feature representation apparatus of the switch point current time series in this embodiment includes a statistical module 501, an extraction module 502, and a merging module 503, where:
the statistical module 501 is configured to perform statistics on a current time sequence of a turnout switch machine to obtain statistical characteristics of the current time sequence;
the current time sequence is a sequence formed by sequencing the current of the turnout switch machine at each moment in a certain historical time period according to the sequence of the acquisition time. The statistical module 501 performs statistics on the current values in the current time series, and takes the obtained statistical characteristics as statistical characteristics. The statistical features include one or more of a current maximum, a current minimum, and a current average.
The extraction module 502 is configured to perform feature extraction on the current time sequence based on a deep neural network, so as to obtain a time sequence feature of the current time sequence;
the present embodiment is not limited to the kind of deep neural network. A timing characteristic is a characteristic that reflects the timing characteristics of a current in a current time series. The extraction module 502 extracts timing features of the current time series using a deep neural network.
The merging module 503 is configured to merge the statistical characteristic and the time sequence characteristic of the current time series, and use a merging result as a characterization of the current time series.
And judging whether the working state of the turnout switch machine is normal or not by using a machine learning model according to the representation of the current time sequence. And if the working state of the turnout switch machine is abnormal, further determining the fault type of the turnout switch machine by using a machine learning model.
In the embodiment, the statistical characteristics are obtained by counting the current time sequence of the turnout switch machine from the global and static angles in the statistical characteristic extraction part; the time sequence feature extraction part extracts the time sequence feature of the current time sequence from a dynamic angle based on a deep neural network, and combines the two part features to be used as the feature representation of the current time sequence of the turnout switch machine, so that the static feature and the dynamic feature are comprehensively used for representing the current time sequence, the relevant characteristics of the current time sequence are better represented, and accordingly, the turnout switch machine can be more accurately detected in an abnormal mode and the fault type of the turnout switch machine can be more accurately identified.
On the basis of the above embodiment, the extraction module in this embodiment includes a conversion sub-module, a coding sub-module, and an extraction sub-module; the conversion submodule is used for converting the current time sequence into a symbol sequence; wherein the symbol sequence is a sequence represented by a plurality of symbols; an encoding submodule for performing a one-hot encoding on the symbol sequence; and the extraction submodule is used for extracting the characteristic of the one-hot code of the symbol sequence based on the deep neural network to obtain the time sequence characteristic of the current time sequence.
On the basis of the foregoing embodiment, the conversion sub-module in this embodiment is specifically configured to: representing the current value of each moment in the current time sequence by a symbol; wherein the same current value is represented by the same symbol, and different current values are represented by different symbols.
On the basis of the foregoing embodiment, the conversion sub-module in this embodiment is specifically configured to: acquiring a current curve of the turnout switch machine according to the current time sequence; dividing the current curve into a plurality of line segments, and calculating the current average value of each line segment; representing each line segment by a symbol according to the current average value of each line segment; wherein the line segments with the same current average value are represented by the same symbol, and the line segments with different current average values are represented by different symbols.
On the basis of the foregoing embodiment, the coding sub-module in this embodiment is specifically configured to: for any symbol in the symbol sequence, setting the state of a state bit corresponding to the symbol in a preset state bit sequence as valid, and setting other state bits except the state bit corresponding to the symbol in the state bit sequence as invalid; wherein each state bit in the state bit sequence corresponds to a symbol.
On the basis of the above embodiment, the extraction sub-module in this embodiment is specifically configured to: for any symbol in the symbol sequence, taking the one-hot coding of the context of the symbol in the one-hot coding of the symbol sequence as the input of the deep neural network, and training the deep neural network so that the output layer of the deep neural network outputs the one-hot coding of the symbol in the one-hot coding of the symbol sequence; taking the output of a hidden layer before the output layer in the trained deep neural network as the vector representation of the symbol; and taking the average value of the vector representation of all the symbols as the time sequence characteristic of the current time sequence.
On the basis of the foregoing embodiment, the extraction sub-module in this embodiment is further configured to: for any symbol in the symbol sequence, taking all other symbols except the symbol in a window which takes the symbol as the center in the symbol sequence as the context of the symbol; wherein the size of the window is preset.
Fig. 6 illustrates a physical structure diagram of an electronic device, which may include, as shown in fig. 6: a processor (processor)601, a communication interface (communication interface)602, a memory (memory)603 and a communication bus 604, wherein the processor 601, the communication interface 602 and the memory 603 complete communication with each other through the communication bus 604. The processor 601 may invoke logic instructions in the memory 603 to perform a method for characterizing a time series of switch machine currents, the method comprising: counting a current time sequence of a turnout switch machine to obtain statistical characteristics of the current time sequence; extracting the characteristics of the current time sequence based on a deep neural network to obtain the time sequence characteristics of the current time sequence; and combining the statistical characteristics and the time sequence characteristics of the current time sequence, and taking the combined result as the representation of the current time sequence.
In addition, the logic instructions in the memory 603 may be implemented in the form of software functional units and stored in a computer readable storage medium when the logic instructions are sold or used as independent products. Based on such understanding, the technical solution of the present invention may be embodied in the form of a software product, which is stored in a storage medium and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: various media capable of storing program codes, such as a usb disk, a removable hard disk, a Read-only memory (ROM), a Random Access Memory (RAM), a magnetic disk, or an optical disk.
In another aspect, an embodiment of the present invention further provides a computer program product, where the computer program product includes a computer program stored on a non-transitory computer-readable storage medium, the computer program includes program instructions, and when the program instructions are executed by a computer, the computer is capable of executing a method for characterizing a time series of switch machine current provided by the above-mentioned embodiments of the methods, where the method includes: counting a current time sequence of a turnout switch machine to obtain statistical characteristics of the current time sequence; extracting the characteristics of the current time sequence based on a deep neural network to obtain the time sequence characteristics of the current time sequence; and combining the statistical characteristics and the time sequence characteristics of the current time sequence, and taking the combined result as the representation of the current time sequence.
In still another aspect, an embodiment of the present invention further provides a non-transitory computer-readable storage medium, on which a computer program is stored, where the computer program is implemented by a processor to execute a method for characterizing a time series of current of a switch machine provided in the foregoing embodiments, where the method includes: counting a current time sequence of a turnout switch machine to obtain statistical characteristics of the current time sequence; extracting the characteristics of the current time sequence based on a deep neural network to obtain the time sequence characteristics of the current time sequence; and combining the statistical characteristics and the time sequence characteristics of the current time sequence, and taking the combined result as the representation of the current time sequence.
The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.
Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. With this understanding in mind, the above-described technical solutions may be embodied in the form of a software product, which can be stored in a computer-readable storage medium such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in the embodiments or some parts of the embodiments.
Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.
Claims (10)
1. A method for characterizing a time series of a switch machine current, comprising:
counting a current time sequence of a turnout switch machine to obtain statistical characteristics of the current time sequence;
extracting the characteristics of the current time sequence based on a deep neural network to obtain the time sequence characteristics of the current time sequence;
and combining the statistical characteristics and the time sequence characteristics of the current time sequence, and taking the combined result as the representation of the current time sequence.
2. The method for characterizing a switch machine current time series according to claim 1, wherein the step of extracting the current time series based on a deep neural network comprises the steps of:
converting the current time series into a symbol series; wherein the symbol sequence is a sequence represented by a plurality of symbols;
performing one-hot encoding on the symbol sequence;
and extracting the characteristic of the one-hot code of the symbol sequence based on the deep neural network to obtain the time sequence characteristic of the current time sequence.
3. The method for characterizing a switch machine current time series of a switch machine as claimed in claim 2, wherein the step of converting the current time series into a symbol sequence comprises:
representing the current value of each moment in the current time sequence by a symbol; wherein the same current value is represented by the same symbol, and different current values are represented by different symbols.
4. The method for characterizing a switch machine current time series of a switch machine as claimed in claim 2, wherein the step of converting the current time series into a symbol sequence comprises:
acquiring a current curve of the turnout switch machine according to the current time sequence;
dividing the current curve into a plurality of line segments, and calculating the current average value of each line segment;
representing each line segment by a symbol according to the current average value of each line segment; wherein the line segments with the same current average value are represented by the same symbol, and the line segments with different current average values are represented by different symbols.
5. The method of characterizing a time series of switch machine currents as claimed in claim 2, wherein said step of encoding said symbol series exclusively comprises:
for any symbol in the symbol sequence, setting the state of a state bit corresponding to the symbol in a preset state bit sequence as valid, and setting other state bits except the state bit corresponding to the symbol in the state bit sequence as invalid; wherein each state bit in the state bit sequence corresponds to a symbol.
6. The method for characterizing a switch machine current time series according to claim 2, wherein the step of obtaining the time series characteristics of the current time series based on the deep neural network performing feature extraction on the one-hot coding of the symbol series comprises:
for any symbol in the symbol sequence, taking the one-hot coding of the context of the symbol in the one-hot coding of the symbol sequence as the input of the deep neural network, and training the deep neural network so that the output layer of the deep neural network outputs the one-hot coding of the symbol in the one-hot coding of the symbol sequence;
taking the output of a hidden layer before the output layer in the trained deep neural network as the vector representation of the symbol;
and taking the average value of the vector representation of all the symbols as the time sequence characteristic of the current time sequence.
7. The method for characterizing a time series of switch machine currents as claimed in claim 6, wherein said step of training said deep neural network further comprises, for any symbol in said sequence of symbols, a one-hot coding of a context of said symbol in said one-hot coding of said sequence of symbols as an input to said deep neural network:
for any symbol in the symbol sequence, taking all other symbols except the symbol in a window which takes the symbol as the center in the symbol sequence as the context of the symbol; wherein the size of the window is preset.
8. A device for characterizing a time series of a current of a switch machine, comprising:
the statistical module is used for carrying out statistics on the current time sequence of the turnout switch machine to obtain the statistical characteristics of the current time sequence;
the extraction module is used for extracting the characteristics of the current time sequence based on a deep neural network to obtain the time sequence characteristics of the current time sequence;
and the merging module is used for merging the statistical characteristics and the time sequence characteristics of the current time sequence and taking the merging result as the representation of the current time sequence.
9. An electronic device comprising a memory, a processor and a computer program stored on the memory and executable on the processor, wherein the processor when executing the program implements the steps of the method for characterizing a time series of switch machine currents according to any one of claims 1 to 7.
10. A non-transitory computer-readable storage medium, on which a computer program is stored, wherein the computer program, when being executed by a processor, implements the steps of the method for characterizing a temporal sequence of current for a switch machine according to any one of claims 1 to 7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010664463.3A CN111881950B (en) | 2020-07-10 | 2020-07-10 | Method and device for representing characteristics of current time sequence of turnout point machine |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202010664463.3A CN111881950B (en) | 2020-07-10 | 2020-07-10 | Method and device for representing characteristics of current time sequence of turnout point machine |
Publications (2)
Publication Number | Publication Date |
---|---|
CN111881950A true CN111881950A (en) | 2020-11-03 |
CN111881950B CN111881950B (en) | 2024-05-21 |
Family
ID=73151199
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202010664463.3A Active CN111881950B (en) | 2020-07-10 | 2020-07-10 | Method and device for representing characteristics of current time sequence of turnout point machine |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN111881950B (en) |
Cited By (3)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112348170A (en) * | 2020-11-10 | 2021-02-09 | 交控科技股份有限公司 | Fault diagnosis method and system for turnout switch machine |
CN116482457A (en) * | 2023-03-24 | 2023-07-25 | 四川众合智控科技有限公司 | Intelligent turnout diagnosis method based on relay interlocking |
CN117741512A (en) * | 2024-02-20 | 2024-03-22 | 山东铁路投资控股集团有限公司 | Switch machine state detection method and system based on neural network |
Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107215357A (en) * | 2017-05-25 | 2017-09-29 | 同济大学 | A kind of switch breakdown Forecasting Methodology |
CN107527091A (en) * | 2016-10-14 | 2017-12-29 | 腾讯科技(北京)有限公司 | Data processing method and device |
CN108879692A (en) * | 2018-06-26 | 2018-11-23 | 湘潭大学 | A kind of regional complex energy resource system energy flow distribution prediction technique and system |
CN109620211A (en) * | 2018-11-01 | 2019-04-16 | 吉林大学珠海学院 | A kind of intelligent abnormal electrocardiogram aided diagnosis method based on deep learning |
US20190197409A1 (en) * | 2017-01-09 | 2019-06-27 | Onu Technology Inc. | Method for unsupervised sequence learning using reinforcement learning and neural networks |
US20190294953A1 (en) * | 2018-03-20 | 2019-09-26 | International Business Machines Corporation | Comparing time series data using context-based similarity |
CN110441629A (en) * | 2019-07-29 | 2019-11-12 | 交控科技股份有限公司 | Method for diagnosing faults and device based on point machine action current curve |
CN110830435A (en) * | 2019-08-27 | 2020-02-21 | 国家电网有限公司信息通信分公司 | Method and device for extracting network flow space-time characteristics and detecting abnormity |
CN110837718A (en) * | 2019-11-07 | 2020-02-25 | 交控科技股份有限公司 | Turnout fault detection method and device, electronic equipment and storage medium |
CN111046942A (en) * | 2019-12-09 | 2020-04-21 | 交控科技股份有限公司 | Turnout fault judgment method and device |
-
2020
- 2020-07-10 CN CN202010664463.3A patent/CN111881950B/en active Active
Patent Citations (10)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN107527091A (en) * | 2016-10-14 | 2017-12-29 | 腾讯科技(北京)有限公司 | Data processing method and device |
US20190197409A1 (en) * | 2017-01-09 | 2019-06-27 | Onu Technology Inc. | Method for unsupervised sequence learning using reinforcement learning and neural networks |
CN107215357A (en) * | 2017-05-25 | 2017-09-29 | 同济大学 | A kind of switch breakdown Forecasting Methodology |
US20190294953A1 (en) * | 2018-03-20 | 2019-09-26 | International Business Machines Corporation | Comparing time series data using context-based similarity |
CN108879692A (en) * | 2018-06-26 | 2018-11-23 | 湘潭大学 | A kind of regional complex energy resource system energy flow distribution prediction technique and system |
CN109620211A (en) * | 2018-11-01 | 2019-04-16 | 吉林大学珠海学院 | A kind of intelligent abnormal electrocardiogram aided diagnosis method based on deep learning |
CN110441629A (en) * | 2019-07-29 | 2019-11-12 | 交控科技股份有限公司 | Method for diagnosing faults and device based on point machine action current curve |
CN110830435A (en) * | 2019-08-27 | 2020-02-21 | 国家电网有限公司信息通信分公司 | Method and device for extracting network flow space-time characteristics and detecting abnormity |
CN110837718A (en) * | 2019-11-07 | 2020-02-25 | 交控科技股份有限公司 | Turnout fault detection method and device, electronic equipment and storage medium |
CN111046942A (en) * | 2019-12-09 | 2020-04-21 | 交控科技股份有限公司 | Turnout fault judgment method and device |
Non-Patent Citations (3)
Title |
---|
HAIQIN YANG ET AL.: "Implicit Heterogeneous Features Embedding in Deep Knowledge Tracing", COGN COMPUT, vol. 10, 15 December 2017 (2017-12-15), XP036440656, DOI: 10.1007/s12559-017-9522-0 * |
张凯;杜凯;巨永锋;: "基于BP神经网络的提速道岔故障诊断算法", 武汉理工大学学报, no. 11, 30 November 2014 (2014-11-30) * |
李俊铭 等: "基于改进IPSO-BP算法的CAM编程切削参数预测", 机电工程技术, vol. 49, no. 5, 31 May 2020 (2020-05-31) * |
Cited By (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN112348170A (en) * | 2020-11-10 | 2021-02-09 | 交控科技股份有限公司 | Fault diagnosis method and system for turnout switch machine |
CN116482457A (en) * | 2023-03-24 | 2023-07-25 | 四川众合智控科技有限公司 | Intelligent turnout diagnosis method based on relay interlocking |
CN116482457B (en) * | 2023-03-24 | 2024-05-14 | 四川众合智控科技有限公司 | Intelligent turnout diagnosis method based on relay interlocking |
CN117741512A (en) * | 2024-02-20 | 2024-03-22 | 山东铁路投资控股集团有限公司 | Switch machine state detection method and system based on neural network |
CN117741512B (en) * | 2024-02-20 | 2024-06-07 | 山东铁路投资控股集团有限公司 | Switch machine state detection method and system based on neural network |
Also Published As
Publication number | Publication date |
---|---|
CN111881950B (en) | 2024-05-21 |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN111881950A (en) | Method and device for representing characteristics of current time sequence of turnout switch machine | |
CN108737406B (en) | Method and system for detecting abnormal flow data | |
CN113853624A (en) | Method and device for predicting power load data and storage medium | |
CN110826525B (en) | Face recognition method and system | |
CN112215409B (en) | Rail transit station passenger flow prediction method and system | |
CN111899759B (en) | Method, device, equipment and medium for pre-training and model training of audio data | |
CN112036505A (en) | Method and device for determining equipment state of turnout switch machine and electronic equipment | |
CN113259216B (en) | CAN bus signal analysis method and system | |
CN107766188B (en) | Memory detection method and device in train control system | |
CN113129870A (en) | Training method, device, equipment and storage medium of speech recognition model | |
CN113420681A (en) | Behavior recognition and model training method, apparatus, storage medium, and program product | |
CN113114529A (en) | KPI (Key Performance indicator) abnormity detection method and device based on condition variation automatic encoder | |
CN111898661A (en) | Method and device for monitoring working state of turnout switch machine | |
CN110971784B (en) | Video processing method and device, electronic equipment and storage medium | |
CN113688887A (en) | Training and image recognition method and device of image recognition model | |
CN115550053A (en) | Monitoring alarm prediction method and device | |
CN114445663A (en) | Method, apparatus and computer program product for detecting challenge samples | |
CN113670611A (en) | Bearing early degradation evaluation method, system, medium and electronic equipment | |
EP2105867B1 (en) | Method and system for line segment extraction | |
CN113378921A (en) | Data screening method and device and electronic equipment | |
CN112348170A (en) | Fault diagnosis method and system for turnout switch machine | |
CN115169650B (en) | Equipment health prediction method for big data analysis | |
CN115964470A (en) | Service life prediction method and system for motorcycle accessories | |
CN114332699B (en) | Road condition prediction method, device, equipment and storage medium | |
CN113223049B (en) | Track data processing method and device |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination | ||
GR01 | Patent grant | ||
GR01 | Patent grant |