CN117520964A - Motor pre-diagnosis method based on artificial intelligence - Google Patents
Motor pre-diagnosis method based on artificial intelligence Download PDFInfo
- Publication number
- CN117520964A CN117520964A CN202410011112.0A CN202410011112A CN117520964A CN 117520964 A CN117520964 A CN 117520964A CN 202410011112 A CN202410011112 A CN 202410011112A CN 117520964 A CN117520964 A CN 117520964A
- Authority
- CN
- China
- Prior art keywords
- sample
- motor
- decision tree
- feature
- fault
- 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
- 238000003745 diagnosis Methods 0.000 title claims abstract description 115
- 238000000034 method Methods 0.000 title claims abstract description 43
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 39
- 238000003066 decision tree Methods 0.000 claims abstract description 150
- 238000012360 testing method Methods 0.000 claims description 71
- 238000012545 processing Methods 0.000 claims description 15
- 238000004590 computer program Methods 0.000 claims description 11
- 238000012512 characterization method Methods 0.000 claims description 9
- 238000010276 construction Methods 0.000 claims description 5
- 238000002405 diagnostic procedure Methods 0.000 claims description 5
- 238000000605 extraction Methods 0.000 claims description 3
- 239000010720 hydraulic oil Substances 0.000 description 21
- 238000010586 diagram Methods 0.000 description 8
- 238000005192 partition Methods 0.000 description 8
- 238000004891 communication Methods 0.000 description 4
- 238000004422 calculation algorithm Methods 0.000 description 3
- 238000004364 calculation method Methods 0.000 description 3
- 239000000356 contaminant Substances 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 239000007788 liquid Substances 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 230000000903 blocking effect Effects 0.000 description 1
- 238000004140 cleaning Methods 0.000 description 1
- 230000003749 cleanliness Effects 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 230000008030 elimination Effects 0.000 description 1
- 238000003379 elimination reaction Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000006870 function Effects 0.000 description 1
- 230000036541 health Effects 0.000 description 1
- 239000012535 impurity Substances 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
- 239000002245 particle Substances 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 239000007787 solid Substances 0.000 description 1
- 238000006467 substitution reaction Methods 0.000 description 1
- 230000000007 visual effect Effects 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
- G06F18/243—Classification techniques relating to the number of classes
- G06F18/24323—Tree-organised classifiers
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F17/00—Digital computing or data processing equipment or methods, specially adapted for specific functions
- G06F17/10—Complex mathematical operations
- G06F17/11—Complex mathematical operations for solving equations, e.g. nonlinear equations, general mathematical optimization problems
-
- 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
- G06F18/213—Feature extraction, e.g. by transforming the feature space; Summarisation; Mappings, e.g. subspace methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N5/00—Computing arrangements using knowledge-based models
- G06N5/01—Dynamic search techniques; Heuristics; Dynamic trees; Branch-and-bound
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F2218/00—Aspects of pattern recognition specially adapted for signal processing
- G06F2218/08—Feature extraction
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- Data Mining & Analysis (AREA)
- General Physics & Mathematics (AREA)
- Mathematical Physics (AREA)
- General Engineering & Computer Science (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Artificial Intelligence (AREA)
- Evolutionary Computation (AREA)
- Software Systems (AREA)
- Mathematical Optimization (AREA)
- Bioinformatics & Computational Biology (AREA)
- Pure & Applied Mathematics (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Biology (AREA)
- Computational Mathematics (AREA)
- Mathematical Analysis (AREA)
- Computing Systems (AREA)
- Operations Research (AREA)
- Computational Linguistics (AREA)
- Algebra (AREA)
- Databases & Information Systems (AREA)
- Testing And Monitoring For Control Systems (AREA)
Abstract
The application relates to the field of fault diagnosis and provides an artificial intelligence-based motor pre-diagnosis method, which comprises the following steps: acquiring operation parameters affecting the state of a motor in the working machine; inputting the operation parameters into a motor diagnosis model, and extracting a plurality of dimension characteristics from the operation parameters by the motor diagnosis model; and traversing a decision path in a feature decision tree based on each dimension feature to obtain a motor diagnosis result of the motor. According to the motor pre-diagnosis method based on artificial intelligence, multiple dimension features are extracted from the operation parameters, so that fault diagnosis can be comprehensively carried out on the feature decision tree based on the multiple dimension features, motor diagnosis results can be accurately and rapidly obtained, and the problem that false alarm or false alarm is easy to occur in the traditional method is avoided.
Description
Technical Field
The application relates to the technical field of fault diagnosis, in particular to a motor pre-diagnosis method based on artificial intelligence.
Background
The motor in the work machine refers to an assembly for filtering impurities and solid particles in a liquid or gas, which is located in a liquid or gas flow path, and functions to filter and remove contaminants to protect the normal operation of the work machine and to extend its life. If the motor is clogged, the filtering effect will be affected, causing contaminants to enter the operating system, increasing the friction and wear risks of the mechanical parts. Therefore, diagnosing whether the motor is clogged or not and replacing the motor in time are key steps in maintaining the health of the work machine.
Currently, whether the motor is jammed is determined according to whether the motor pressure is greater than a set threshold value or not and the number of times the threshold value occurs in a set period of time. However, when a severe working condition is met, the alarm condition can be met in a short period, the motor is not blocked in practice, and false alarm is easy to occur; in addition, when the set threshold value is larger, the motor is blocked, so that an alarm is missed.
Disclosure of Invention
The application provides a motor pre-diagnosis method based on artificial intelligence, which is used for solving the defect that the motor pre-diagnosis based on artificial intelligence in the prior art is easy to cause false alarm or missing alarm.
In a first aspect, the present application provides an artificial intelligence based motor pre-diagnosis method, comprising:
acquiring operation parameters affecting the state of a motor in the working machine;
inputting the operation parameters into a motor diagnosis model, and extracting a plurality of dimension characteristics from the operation parameters by the motor diagnosis model; the plurality of dimension features are used for representing the attributes of different dimensions of the operation parameters; the motor diagnosis model is constructed based on a feature decision tree, and the feature decision tree is constructed based on a plurality of sample dimension features corresponding to sample operation parameters and sample fault labels; the sample fault tag is used for representing the state of the motor corresponding to the sample operation parameter; the motor state includes a blocked state and an unblocked state;
And traversing a decision path in a feature decision tree based on the dimension features to obtain a motor diagnosis result of the motor.
In an embodiment, performing decision path traversal in a feature decision tree based on the dimensional features to obtain a motor diagnosis result of the motor, including:
taking a root node of the feature decision tree as a first starting node, and selecting a path corresponding to the dimension feature value along a branch of the feature decision tree to obtain a child node of the root node;
selecting a path of corresponding dimension characteristic values along branches of the characteristic decision tree by taking the child node as a second starting node until the child node reaches leaf nodes of the characteristic decision tree;
and determining the fault state corresponding to the leaf node as the motor diagnosis result.
In one embodiment, the step of constructing the feature decision tree includes:
dividing: traversing each characteristic attribute corresponding to a current sample set, and dividing the current sample set based on each characteristic attribute of the current sample set to obtain a current sample sub-set corresponding to each characteristic attribute; the current sample set comprises all sample dimension characteristics corresponding to the sample operation parameters;
Determining: determining information gain corresponding to the corresponding characteristic attribute based on the duty ratio of each sample fault label in each current sample subset;
the construction steps are as follows: taking a characteristic attribute corresponding to the maximum information gain as a dividing attribute of a current node of the characteristic decision tree, taking a current sample subset corresponding to the dividing attribute as the current sample set, and deleting sample dimension characteristics corresponding to the dividing attribute from the current sample set;
and returning to sequentially and repeatedly executing the dividing step, the determining step and the constructing step until no sample dimension characteristic exists in the current sample set.
In an embodiment, determining the information gain corresponding to the corresponding feature attribute based on the duty ratio of the sample dimension feature corresponding to each sample failure label in each current sample subset includes:
determining a first information entropy based on the duty ratio of various sample fault labels in each current sample subset; the first information entropy representation divides the current sample set according to the characteristic attribute to obtain an average value of data purity of each current sample subset; the data purity of each current sample subset is determined based on the proportion of each sample fault label in each current sample subset;
Determining a second information entropy based on the duty ratio of various sample fault labels in the current sample set; the second information entropy represents the data purity of the current sample set based on the proportion of various sample fault labels in the current sample set;
and determining the information gain corresponding to the corresponding characteristic attribute based on the first information entropy and the second information entropy.
In one embodiment, after the feature decision tree is constructed, the method further comprises:
acquiring a plurality of test dimension characteristics corresponding to test operation parameters and test fault labels corresponding to the test operation parameters; the test operation parameters represent operation parameters for testing the precision of the characteristic decision tree; the test dimension features are used for representing the attributes of different dimensions of the operation parameters;
applying each test dimension characteristic based on the characteristic decision tree, and determining a test fault result;
determining the precision of the feature decision tree based on the test fault result and the test fault label;
and under the condition that the precision of the feature decision tree is smaller than a threshold value, a plurality of sample dimension features and sample fault labels corresponding to the sample operation parameters are obtained in an increment mode, and the feature decision tree is updated based on the plurality of sample dimension features and the sample fault labels obtained in the increment mode.
In an embodiment, determining the accuracy of the feature decision tree based on the test fault result and the test fault label comprises:
determining a real example, a false positive example, a true counterexample and a false counterexample based on the test fault result and the test fault label; the real case characterization correctly predicts the fault sample as a fault sample; the false positive token incorrectly predicts a non-faulty sample as a faulty sample; the true counterexample characterization correctly predicts a non-faulty sample as a non-faulty sample; the false counter characterization incorrectly predicts a failed sample as a non-failed sample;
determining a recall ratio and an precision ratio based on the true example, the false positive example, the true negative example and the false negative example; the recall ratio characterizes the recognition capability of the feature decision tree to a fault sample; the precision represents the accuracy of the feature decision tree in a sample predicted to be a fault;
and determining the precision of the feature decision tree based on the recall ratio and the precision ratio.
In an embodiment, before constructing the feature decision tree based on the plurality of sample dimension features and the sample fault labels corresponding to the sample operation parameters, the method further includes:
Performing data processing on the sample operation parameters to obtain sample operation parameters after data processing;
extracting a plurality of initial dimension features from the sample operation parameters after data processing; the plurality of initial dimensional features includes time domain dimensional features and frequency domain dimensional features;
and carrying out feature selection on each initial dimension feature to obtain the plurality of sample dimension features.
In a second aspect, the present application further provides an artificial intelligence based motor pre-diagnosis apparatus, including:
an acquisition unit for acquiring an operation parameter affecting a state of a motor in the work machine;
an extraction unit for inputting the operation parameters to a motor diagnosis model, and extracting a plurality of dimension features from the operation parameters by the motor diagnosis model; the plurality of dimension features are used for representing the attributes of different dimensions of the operation parameters; the motor diagnosis model is constructed based on a feature decision tree, and the feature decision tree is constructed based on a plurality of sample dimension features corresponding to sample operation parameters and sample fault labels; the sample fault tag is used for representing the state of the motor corresponding to the sample operation parameter; the motor state includes a blocked state and an unblocked state;
And the diagnosis unit is used for traversing a decision path in the feature decision tree based on the dimension features to obtain a motor diagnosis result of the motor.
In a third aspect, the present application further provides an electronic device, including a memory, a processor, and a computer program stored on the memory and executable on the processor, where the processor implements the artificial intelligence based motor pre-diagnosis method according to any one of the first aspects above when the computer program is executed.
In a fourth aspect, the present application also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, implements an artificial intelligence based motor pre-diagnostic method as described in any of the first aspects above.
In a fifth aspect, the present application also provides a computer program product comprising a computer program which, when executed by a processor, implements an artificial intelligence based motor pre-diagnosis method as described in any of the first aspects above.
According to the motor pre-diagnosis method based on artificial intelligence, multiple dimension features are extracted from the operation parameters, so that fault diagnosis can be comprehensively carried out on the feature decision tree based on the multiple dimension features, motor diagnosis results can be accurately and rapidly obtained, and the problem that false alarm or false alarm is easy to occur in the traditional method is avoided.
Drawings
For a clearer description of the present application or of the prior art, the drawings that are used in the description of the embodiments or of the prior art will be briefly described, it being apparent that the drawings in the description below are some embodiments of the present application, and that other drawings may be obtained from these drawings without inventive effort for a person skilled in the art.
FIG. 1 is one of the flow diagrams of the artificial intelligence based motor pre-diagnostic method provided herein;
FIG. 2 is a schematic diagram of a feature decision tree provided herein;
FIG. 3 is a flow diagram of feature selection provided herein;
FIG. 4 is a second schematic flow chart of the motor pre-diagnosis method based on artificial intelligence provided by the application;
FIG. 5 is a schematic diagram of the structure of an artificial intelligence based motor pre-diagnostic device provided herein;
fig. 6 is a schematic structural diagram of an electronic device provided in the present application.
Detailed Description
For the purposes of making the objects, technical solutions and advantages of the present application more apparent, the technical solutions in the present application will be clearly and completely described below with reference to the drawings in the present application, and it is apparent that the described embodiments are some, but not all, embodiments of the present application. All other embodiments, which can be made by one of ordinary skill in the art based on the embodiments herein without making any inventive effort, are intended to be within the scope of the present application.
The application provides a motor pre-diagnosis method based on artificial intelligence. FIG. 1 is a schematic flow chart of an artificial intelligence-based motor pre-diagnosis method provided by the application, as shown in FIG. 1, the method comprises the following steps:
step 110, obtaining an operating parameter that affects a state of a motor in the work machine.
Here, the motor status refers to the current operating condition and performance of the motor, and is used to describe the degree of cleanliness, clogging or damage of the motor and whether maintenance, replacement or cleaning of the motor is required. The work machine in the present application may be an excavator, a crane, a loader, or the like.
The operation parameters refer to parameters affecting the state of the motor during operation of the work machine, and may include motor pressure, hydraulic oil temperature, main pump pressure, and the like. Alternatively, operating parameters affecting the state of the motor in the work machine may be obtained in real time via sensors.
Step 120, the operation parameters are input into a motor diagnosis model, and the motor diagnosis model extracts a plurality of dimension features from the operation parameters.
In particular, a plurality of dimensional features are used to characterize the properties of different dimensions of the operating parameters. For example, the operation parameters are motor pressure, and the maximum value, the minimum value, the 1/4 quantile, the 3/4 quantile, the average value and the like of the motor pressure can be extracted from the time domain dimension, and the frequency width average value of the motor pressure in the frequency range of 0-10Hz, the frequency width average value of the motor pressure in the frequency range of 10-20Hz, the frequency width average value of the motor pressure in the frequency range of 20-30Hz, the frequency width average value of the motor pressure in the frequency range of 30-40Hz, the frequency width average value of the motor pressure in the frequency range of 40-50Hz, the frequency width average value of the motor pressure in the frequency range of 50-70Hz, the frequency width average value of the motor pressure in the frequency range of 70-100Hz and the like can be extracted from the frequency domain dimension. The multidimensional features can provide more comprehensive and more accurate data description, and hidden association relations are found, so that motor diagnosis results of the motor can be determined more accurately based on the multidimensional features.
In addition, the motor diagnosis model is constructed based on a feature decision tree, and the feature decision tree is constructed based on a plurality of sample dimension features corresponding to the sample operation parameters and sample fault labels. The sample fault tag is used for representing that the sample operation parameter corresponds to a motor state, and the corresponding motor state comprises a blocking state and an unblocking state.
And 130, traversing a decision path in a feature decision tree based on each dimension feature to obtain a motor diagnosis result of the motor.
In the feature decision tree, the decision path is a path for selecting a corresponding dimension feature value along a branch of the decision tree from a root node until reaching a leaf node, and then taking a fault state identified by the corresponding leaf node as a motor diagnosis result.
For example, the root node of the feature decision tree is feature a, and there are two branches, and the values corresponding to feature a are yes and no, respectively. If the "yes" branch of feature A is followed, then the next node is entered, which is feature B, and the selection of feature B's value continues. And so on until the leaf node is reached.
In addition, the structure of the feature decision tree is visual, so that the feature decision tree is easy to understand and explain, and by looking at the decision path, the motor diagnosis model can be known to predict the motor diagnosis result. The feature decision tree can also process a plurality of features at the same time, and the importance of each feature can be measured through branches and leaf nodes of the decision tree, namely, the feature decision tree can comprehensively perform fault diagnosis based on the plurality of features, and a motor diagnosis result can be accurately obtained. Furthermore, the prediction speed of the feature decision tree is high, and a large amount of computing resources are not needed.
According to the motor pre-diagnosis method based on the artificial intelligence, multiple dimension features are extracted from the operation parameters, so that fault diagnosis can be comprehensively carried out on the feature decision tree based on the multiple dimension features, motor diagnosis results can be accurately and rapidly obtained, and the problem that false alarm or missing alarm is easy to occur in the traditional method is avoided.
Based on the above embodiment, performing decision path traversal in a feature decision tree based on the dimensional features to obtain a motor diagnosis result of the motor, including:
taking a root node of the feature decision tree as a first starting node, and selecting a path corresponding to the dimension feature value along a branch of the feature decision tree to obtain a child node of the root node;
selecting a path of corresponding dimension characteristic values along branches of the characteristic decision tree by taking the child node as a second starting node until the child node reaches leaf nodes of the characteristic decision tree;
and determining the fault state corresponding to the leaf node as the motor diagnosis result.
Specifically, taking a root node of a feature decision tree as a first starting node, and selecting a path corresponding to a dimension feature value along a branch of the feature decision tree to obtain a child node of the root node. Further, taking the child node as a second starting node, selecting a path corresponding to the dimension characteristic value along the branch of the characteristic decision tree until the path reaches the leaf node of the characteristic decision tree, and determining the fault state corresponding to the leaf node as a motor diagnosis result.
For example, the root node of the feature decision tree is feature a, and there are two branches, and the values corresponding to feature a are yes and no, respectively. And taking the root node of the feature A as a first starting node, selecting a path of the corresponding dimension feature value along the branch of the feature decision tree, and if the branch of the feature A with the value of 'yes' is followed, entering the next node, wherein the node is the feature B. And taking the node of the feature B as a second initial node, selecting a path corresponding to the dimension feature value along the branch of the feature decision tree, and if the path of the feature A with the value of 'Yes' is followed, entering the next node, wherein the node is the feature C, and selecting the path corresponding to the dimension feature value along the branch of the feature decision tree, namely continuously selecting the feature C value. And the like until the leaf node is reached, if the leaf node is the feature M, determining the fault state corresponding to the feature M as a motor diagnosis result.
According to the embodiment of the invention, the plurality of dimension features are extracted from the operation parameters, so that the feature decision tree can comprehensively perform fault diagnosis based on the plurality of dimension features, a motor diagnosis result is accurately and rapidly obtained, and the problem that the traditional method is prone to missing report or false report is avoided.
Based on the above embodiment, the feature decision tree construction steps include:
dividing: traversing each characteristic attribute corresponding to the current sample set, and dividing the current sample set based on each characteristic attribute of the current sample set to obtain a current sample subset corresponding to each characteristic attribute; and the current sample set comprises all sample dimension characteristics corresponding to the sample operation parameters.
Specifically, the feature attributes are used to characterize the class of dimensional features of each sample. If the sample dimension feature 1 is motor pressure 3bar and the sample dimension feature 2 is motor pressure 5.5bar, the sample dimension feature 1 and the sample dimension feature 2 correspond to the same feature attribute, namely the feature attribute is motor pressure. For another example, if the sample dimension feature 3 is the hydraulic oil temperature of 70 ℃, the sample dimension feature 1 and the sample dimension feature 3 correspond to different feature attributes, that is, the feature attribute corresponding to the sample dimension feature 1 is the motor pressure, and the feature attribute corresponding to the sample dimension feature 3 is the hydraulic oil temperature.
In addition, the feature decision tree includes a root node, an internal node, and a leaf node. Firstly, constructing a root node based on an initial sample set, namely taking the initial sample set as a current sample set, traversing characteristic attributes (such as hydraulic oil temperature and motor pressure) corresponding to the current sample set, and dividing the current sample set based on each characteristic attribute of the current sample set to obtain a current sample subset corresponding to each characteristic attribute.
For example, if the characteristic attribute corresponding to the current sample set includes hydraulic oil and motor pressure, the current sample combination may be divided based on the hydraulic oil and the motor pressure, respectively, to obtain a corresponding current sample subset. When the division is based on the hydraulic oil temperature, the current sample set may be divided into two current sample subsets of "hydraulic oil temperature >70 ℃ and" hydraulic oil temperature is less than or equal to 70 ℃. When the division is based on the motor pressure, the current sample set may be divided into "motor pressure >3bar" and "motor pressure ∈3bar".
It should be noted that, the above is to divide the current sample set into two current sample subsets by way of example, but the present sample set is not limited to be divided into two current sample subsets, and may be divided into a plurality of current sample subsets (e.g., three current sample subsets, four current sample subsets, etc.) according to practical situations.
Determining: and determining the information gain corresponding to the corresponding characteristic attribute based on the duty ratio of each sample fault label in each current sample subset.
Specifically, the information gain is used for measuring the importance of the feature attribute to the classification task of the feature decision tree, the larger the information gain is, the more important the corresponding feature attribute is, and the larger the probability that the current sample set is partitioned by taking the corresponding feature attribute as the partition attribute is. The duty ratio of each type of sample fault label in each current sample subset refers to the duty ratio of the corresponding sample dimension feature of each type of sample fault label in all sample dimension features in the current sample subset. For motor pre-diagnostics based on artificial intelligence, sample fault signatures can generally be categorized into two categories, blocked and unblocked.
The construction steps are as follows: taking the characteristic attribute corresponding to the maximum information gain as the dividing attribute of the current node of the characteristic decision tree, taking the current sample subset corresponding to the dividing attribute as the current sample set, and deleting the sample dimension characteristic corresponding to the dividing attribute from the current sample set.
And returning to sequentially and repeatedly executing the dividing step, the determining step and the constructing step until no sample dimension characteristic exists in the current sample set.
Specifically, the larger the information gain, the more important the corresponding feature attribute is in classifying the feature decision tree, and the greater the probability of the corresponding feature attribute as a classification attribute. In this regard, in the embodiment of the present application, the feature attribute corresponding to the maximum information gain is used as the partition attribute of the current node of the feature decision tree, so that the current node can be constructed based on the partition attribute.
And the current sample subset corresponding to the dividing attribute is used as a current sample set, and the sample dimension characteristic corresponding to the dividing attribute is deleted from the current sample set, so that the dividing step, the determining step and the constructing step can be sequentially and repeatedly executed to determine the characteristic attribute of the next node of the characteristic decision tree until the sample dimension characteristic does not exist in the current sample set.
Fig. 2 is a schematic structural diagram of a feature decision tree provided in the present application, as shown in fig. 2, at a root node of the feature decision tree, a partition attribute is a "hydraulic oil temperature average value", at an internal node corresponding to the root node of the feature decision tree, the partition attribute includes a "motor pressure average value" and a "motor pressure 0-10HZ frequency band amplitude average value", a partition attribute of a next internal node corresponding to the "motor pressure average value" is a "motor pressure maximum value", a partition attribute of a next internal node corresponding to the "motor pressure 0-10HZ frequency band amplitude average value" is a "motor pressure 10-20HZ frequency band amplitude average value", and a leaf node of the feature decision tree corresponds to a motor diagnosis result.
Further, determining the information gain corresponding to the corresponding feature attribute based on the duty ratio of the sample dimension feature corresponding to each sample fault label in each current sample subset includes:
determining a first information entropy based on the duty ratio of various sample fault labels in each current sample subset;
determining a second information entropy based on the duty ratio of various sample fault labels in the current sample set;
and determining the information gain corresponding to the corresponding characteristic attribute based on the first information entropy and the second information entropy.
Specifically, the first information entropy is used for representing an average value of data purity of each obtained current sample subset after the current sample set is divided according to the characteristic attribute. The data purity of each current sample subset is determined based on the proportion of each sample fault label in each current sample subset. The second information entropy is used for representing the data purity of the current sample set and is based on the proportion of various sample fault labels in the current sample set.
The proportion of the fault labels of various samples is close, which means that the lower the data purity is, the greater the degree of data confusion is, and the greater the corresponding information entropy is. When the proportion of the fault labels of a certain type of sample is obviously higher than that of the fault labels of other types of sample, the higher the data purity of the current sample set is, namely the lower the data confusion degree is, and the corresponding second information entropy is smaller.
The information gain is used for measuring the amount of information which can be obtained after selecting a certain characteristic attribute for division, namely, the degree of information entropy reduction is shown due to the introduction of the characteristic attribute after dividing the current sample set into different current sample subsets on the current node. The larger the information gain is, the larger the obtained information quantity is, namely, the higher the purity of the data of the current sample subset obtained after division is. Optionally, the information gain = second information entropy-first information entropy.
Wherein the information gain is determined based on the following formula:
；
；
wherein,representing feature attributesIs used for the information gain of (a),the second information entropy is represented by a second information entropy,representing the current set of samples,represent the firstThe duty cycle of the class sample fault label in the current sample set,representing the total number of categories of specimen failure tags in the current specimen set,the first information entropy is represented as such,representing a subset of the current samples,representing the number of current subsets of samples.
Embodiments of the present application aim to find a solution that enables an objective functionThe largest characteristic attribute is taken as the partition attribute.
Based on any of the above embodiments, after constructing the feature decision tree, the method further includes:
acquiring a plurality of test dimension characteristics corresponding to the test operation parameters and test fault labels corresponding to the test operation parameters;
based on the feature decision tree, applying each test dimension feature to determine a test fault result;
determining the precision of a feature decision tree based on the test fault result and the test fault label;
and under the condition that the precision of the feature decision tree is smaller than a threshold value, a plurality of sample dimension features and sample fault labels corresponding to the sample operation parameters are obtained in an increment mode, and the feature decision tree is updated based on the plurality of sample dimension features and the sample fault labels obtained in the increment mode.
In particular, test operating parameters refer to operating parameters for testing the accuracy of a feature decision tree, which can be understood as a test set, with test dimension features used to characterize the attributes of the operating parameters in different dimensions. After the feature decision tree is constructed, based on the feature decision tree, each test dimension feature is applied to determine a test fault result, and based on the test fault result and the test fault label, the precision of the feature decision tree is determined. The larger the difference between the test fault result and the test fault label is, the lower the accuracy of the feature decision tree is indicated. Conversely, the smaller the difference between the test fault result and the test fault label, the higher the accuracy of the feature decision tree.
And under the condition that the precision of the feature decision tree is smaller than the threshold value, the feature decision tree is lower in precision, a plurality of sample dimension features and sample fault labels corresponding to the sample operation parameters are obtained in an increment mode, and the feature decision tree is updated based on the plurality of sample dimension features and the sample fault labels which are obtained in the increment mode. When the feature decision tree is updated, the feature decision tree can be updated by adopting the method for constructing the feature decision tree. For example, a plurality of sample dimension features obtained in an incremental manner may be added to the current sample set to update the current sample set, and the step of constructing the feature decision tree is performed to obtain an updated feature decision tree.
Based on any of the above embodiments, determining the accuracy of the feature decision tree based on the test fault result and the test fault label includes:
determining a real example, a false positive example, a true negative example and a false negative example based on the test fault result and the test fault label;
determining a recall ratio and an accuracy ratio based on the true example, the false positive example, the true negative example and the false negative example;
and determining the precision of the feature decision tree based on the recall ratio and the precision ratio.
Specifically, true Positive (TP) refers to correctly predicting a failure sample as a failure sample. False Positive (FP) refers to the False Positive that a non-faulty sample is erroneously predicted as a faulty sample. True Negative (TN) refers to correctly predicting a non-faulty sample as a non-faulty sample. False Negative (FN) refers to a False positive that erroneously predicts a faulty sample as a non-faulty sample.
Recall (Recall), also known as Recall, represents the ability of the feature decision tree to identify fault samples, and its calculation formula may be recall=tp/(tp+fn). The Precision (Precision) represents the accuracy of the feature decision tree in the samples predicted to be faulty, and the calculation formula may be precision=tp/(tp+fp).
The Precision of the feature decision tree may be determined by the Precision and Recall, and the commonly used index may include an F1 score (F1-score), where the F1-score comprehensively considers the Precision and Recall, and is a harmonic mean value of the Precision and Recall, and the F1-score calculation formula may be F1-score=2 x (Precision x Recall)/(precision+recall). The larger the F1-score is, the better the characteristic decision tree is in precision and recall ratio, so that a more accurate motor diagnosis result can be provided.
Based on any of the above embodiments, determining a motor diagnostic result of the motor further comprises:
when the hydraulic oil temperature of the working machine is greater than the threshold value, the motor diagnosis result is pushed to a display device of the working machine.
Considering the difference of hydraulic oil temperatures, the viscosity of hydraulic oil is greatly different, and the motor diagnosis result is greatly influenced. In other words, under the condition that the hydraulic oil temperature is less than or equal to the threshold value, the motor diagnosis result determined based on the feature decision tree is low in accuracy, so that the motor diagnosis result is pushed to the display device of the working machine after the hydraulic oil temperature is greater than the threshold value, and otherwise, the motor diagnosis result is not pushed. Similarly, the hydraulic oil temperature in the sample operation parameters acquired by constructing the feature decision tree is also greater than the threshold value. Alternatively, the motor diagnostic result may be pushed to the display device via the CAN bus
Based on any of the above embodiments, further comprising: before constructing the feature decision tree based on the plurality of sample dimension features and the sample fault labels corresponding to the sample operation parameters, the method further comprises:
performing data processing on the sample operation parameters to obtain sample operation parameters after data processing;
extracting a plurality of initial dimension features from the sample operation parameters after data processing; the plurality of initial dimensional features includes time domain dimensional features and frequency domain dimensional features;
and carrying out feature selection on each initial dimension feature to obtain the plurality of sample dimension features.
Specifically, the database processing may include synchronization of time axes, elimination of null and outliers in the data, and the like. And after the sample operation parameters are subjected to data processing, extracting a plurality of initial dimension features from the sample operation parameters subjected to data processing, wherein the plurality of initial dimension features comprise time domain dimension features and frequency domain dimension features.
For example, in the case where the sample operating parameters include hydraulic oil temperature and motor pressure, the following steps may be employed to extract a plurality of initial dimensional features: dividing a sample operation parameter into a time period of 1s by utilizing a sliding window method, and extracting the maximum value, the minimum value, the 1/4 fraction, the 3/4 fraction and the average value of the motor pressure and the average value of the hydraulic oil temperature in the time period for the time domain dimension; for the frequency domain dimension, a Fourier transform algorithm is utilized to extract the frequency width mean value of the frequency band of 0-10Hz, the frequency width mean value of the frequency band of 10-20Hz, the frequency width mean value of the frequency band of 20-30Hz, the frequency width mean value of the frequency band of 30-40Hz, the frequency width mean value of the frequency band of 40-50Hz, the frequency width mean value of the frequency band of 50-70Hz and the frequency width mean value of the frequency band of 70-100 Hz.
After extracting a plurality of initial dimension features, carrying out feature selection on each initial dimension feature to obtain a plurality of sample dimension features. Wherein feature selection may be performed on each initial dimension feature based on a wrapped feature selection algorithm (Las Vegas Wrapper, LVW) to select an optimal sample dimension feature. Fig. 3 is a schematic flow chart of feature selection provided in the present application, as shown in fig. 3, after feature selection is performed on a plurality of initial dimension features (such as peak value, average value, frequency range average value, etc.) in the above example by adopting a LVW algorithm, the obtained sample dimension features include a hydraulic oil temperature average value, a motor pressure maximum value, a motor pressure average value, a frequency range average value of 0-10Hz frequency range, and a frequency range average value of 10-20Hz frequency range.
Based on any of the above embodiments, fig. 4 is a second schematic flow chart of the motor pre-diagnosis method based on artificial intelligence, and as shown in fig. 4, the operation parameters of the working machine to be diagnosed are obtained in real time, where the operation parameters include motor pressure and hydraulic oil temperature. And then, extracting the characteristics of the operation parameters to obtain a plurality of dimension characteristics, inputting the plurality of dimension characteristics into a motor diagnosis model, determining a motor diagnosis result by the motor diagnosis model based on a characteristic decision tree, and pushing the motor diagnosis result to a display device of the working machine. Wherein the motor diagnostic model is embedded in a controller in the work machine and the hydraulic oil temperature is greater than a threshold value.
In addition, a feature decision tree in the motor diagnosis model is constructed based on a plurality of sample dimension features and sample fault labels corresponding to sample operation parameters, and the specific steps are as follows:
firstly, carrying out feature engineering (feature extraction and feature selection) on sample operation parameters to obtain a plurality of sample dimension features;
then, a feature decision tree is constructed based on the plurality of sample dimension features and the sample fault labels corresponding to the sample operation parameters by adopting the method of the embodiment.
After the feature decision tree is constructed, determining a test fault result based on the application of each test dimension feature; determining the precision (such as F1-score) of the feature decision tree based on the test fault result and the test fault label; and when the precision of the feature decision tree is smaller than a threshold value, a plurality of sample dimension features and sample fault labels corresponding to the sample operation parameters are obtained in an increment mode, and the feature decision tree is updated based on the plurality of sample dimension features and the sample fault labels obtained in the increment mode.
The motor pre-diagnosis device based on artificial intelligence provided by the application is described below, and the motor pre-diagnosis device based on artificial intelligence described below and the motor pre-diagnosis method based on artificial intelligence described above can be referred to correspondingly.
Based on any of the above embodiments, fig. 5 is a schematic structural diagram of an artificial intelligence-based motor pre-diagnosis apparatus provided in the present application, as shown in fig. 5, the apparatus includes:
an acquisition unit 510 for acquiring an operation parameter affecting a motor state in the work machine;
an extracting unit 520 for inputting the operation parameters to a motor diagnosis model, and extracting a plurality of dimensional features from the operation parameters by the motor diagnosis model; the plurality of dimension features are used for representing the attributes of different dimensions of the operation parameters; the motor diagnosis model is constructed based on a feature decision tree, and the feature decision tree is constructed based on a plurality of sample dimension features corresponding to sample operation parameters and sample fault labels; the sample fault tag is used for representing the state of the motor corresponding to the sample operation parameter; the motor state includes a blocked state and an unblocked state;
and the diagnosis unit 530 is configured to perform decision path traversal in a feature decision tree based on the dimensional features, so as to obtain a motor diagnosis result of the motor.
In an embodiment, the diagnostic unit 530 is further configured to:
taking a root node of the feature decision tree as a first starting node, and selecting a path corresponding to the dimension feature value along a branch of the feature decision tree to obtain a child node of the root node;
Selecting a path of corresponding dimension characteristic values along branches of the characteristic decision tree by taking the child node as a second starting node until the child node reaches leaf nodes of the characteristic decision tree;
and determining the fault state corresponding to the leaf node as the motor diagnosis result.
In an embodiment, the motor pre-diagnosis device is further configured to:
dividing: traversing each characteristic attribute corresponding to the current sample set, and dividing the current sample set based on each characteristic attribute of the current sample set to obtain a current sample subset corresponding to each characteristic attribute; the current sample set comprises all sample dimension characteristics corresponding to sample operation parameters;
determining: determining information gain corresponding to the corresponding characteristic attribute based on the duty ratio of each sample fault label in each current sample subset;
the construction steps are as follows: taking the characteristic attribute corresponding to the maximum information gain as the dividing attribute of the current node of the characteristic decision tree, taking the current sample subset corresponding to the dividing attribute as the current sample set, and deleting the sample dimension characteristic corresponding to the dividing attribute from the current sample set;
and returning to sequentially and repeatedly executing the dividing step, the determining step and the constructing step until no sample dimension characteristic exists in the current sample set.
In an embodiment, the motor pre-diagnosis device is further configured to:
determining a first information entropy based on the duty ratio of various sample fault labels in each current sample subset; the first information entropy representation divides the current sample set according to the characteristic attribute to obtain an average value of data purity of each current sample subset; the data purity of each current sample subset is determined based on the proportion of each sample fault label in each current sample subset;
determining a second information entropy based on the duty ratio of various sample fault labels in the current sample set; the second information entropy represents the data purity of the current sample set based on the proportion of various sample fault labels in the current sample set;
and determining the information gain corresponding to the corresponding characteristic attribute based on the first information entropy and the second information entropy.
In an embodiment, the motor pre-diagnosis device is further configured to:
acquiring a plurality of test dimension characteristics corresponding to test operation parameters and test fault labels corresponding to the test operation parameters; the test operation parameters represent operation parameters for testing the precision of the characteristic decision tree; the test dimension features are used for representing the attributes of different dimensions of the operation parameters;
Applying each test dimension characteristic based on the characteristic decision tree, and determining a test fault result;
determining the precision of the feature decision tree based on the test fault result and the test fault label;
and under the condition that the precision of the feature decision tree is smaller than a threshold value, a plurality of sample dimension features and sample fault labels corresponding to the sample operation parameters are obtained in an increment mode, and the feature decision tree is updated based on the plurality of sample dimension features and the sample fault labels obtained in the increment mode.
In an embodiment, the motor pre-diagnosis device is further configured to:
determining a real example, a false positive example, a true counterexample and a false counterexample based on the test fault result and the test fault label; the real case characterization correctly predicts the fault sample as a fault sample; the false positive token incorrectly predicts a non-faulty sample as a faulty sample; the true counterexample characterization correctly predicts a non-faulty sample as a non-faulty sample; the false counter characterization incorrectly predicts a failed sample as a non-failed sample;
determining a recall ratio and an precision ratio based on the true example, the false positive example, the true negative example and the false negative example; the recall ratio characterizes the recognition capability of the feature decision tree to a fault sample; the precision represents the accuracy of the feature decision tree in a sample predicted to be a fault;
And determining the precision of the feature decision tree based on the recall ratio and the precision ratio.
In an embodiment, the motor pre-diagnosis device is further configured to:
performing data processing on the sample operation parameters to obtain sample operation parameters after data processing;
extracting a plurality of initial dimension features from the sample operation parameters after data processing; the plurality of initial dimensional features includes time domain dimensional features and frequency domain dimensional features;
and carrying out feature selection on each initial dimension feature to obtain the plurality of sample dimension features.
Based on any of the above embodiments, the present application further provides a work machine, including: the artificial intelligence based motor pre-diagnostic device as described in any one of the embodiments above. The work machine may be an excavator, a crane, a loader, or the like.
Fig. 6 is a schematic structural diagram of an electronic device provided in the present application, and as shown in fig. 6, the electronic device may include: processor 610, communication interface 620, memory 630, and communication bus 640, wherein processor 610, communication interface 620, and memory 630 communicate with each other via communication bus 640. The processor 610 may invoke logic instructions in the memory 630 to perform an artificial intelligence based motor pre-diagnostic method comprising:
Acquiring operation parameters affecting the state of a motor in the working machine;
inputting the operation parameters into a motor diagnosis model, and extracting a plurality of dimension characteristics from the operation parameters by the motor diagnosis model; the plurality of dimension features are used for representing the attributes of different dimensions of the operation parameters; the motor diagnosis model is constructed based on a feature decision tree, and the feature decision tree is constructed based on a plurality of sample dimension features corresponding to sample operation parameters and sample fault labels; the sample fault tag is used for representing the state of the motor corresponding to the sample operation parameter; the motor state includes a blocked state and an unblocked state;
and traversing a decision path in a feature decision tree based on the dimension features to obtain a motor diagnosis result of the motor.
Further, the logic instructions in the memory 630 may be implemented in the form of software functional units and stored in a computer-readable storage medium when sold or used as a stand-alone product. Based on such understanding, the technical solution of the present application may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in the embodiments of the present application. And the aforementioned storage medium includes: a usb disk, a removable hard disk, a read-only memory (ROM, readOnlyMemory), a random access memory (RAM, randomAccessMemory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
In another aspect, the present application also provides a computer program product comprising a computer program stored on a non-transitory computer readable storage medium, the computer program comprising program instructions which, when executed by a computer, enable the computer to perform the artificial intelligence based motor pre-diagnosis method provided by the above methods, the method comprising:
acquiring operation parameters affecting the state of a motor in the working machine;
inputting the operation parameters into a motor diagnosis model, and extracting a plurality of dimension characteristics from the operation parameters by the motor diagnosis model; the plurality of dimension features are used for representing the attributes of different dimensions of the operation parameters; the motor diagnosis model is constructed based on a feature decision tree, and the feature decision tree is constructed based on a plurality of sample dimension features corresponding to sample operation parameters and sample fault labels; the sample fault tag is used for representing the state of the motor corresponding to the sample operation parameter; the motor state includes a blocked state and an unblocked state;
and traversing a decision path in a feature decision tree based on the dimension features to obtain a motor diagnosis result of the motor.
In yet another aspect, the present application also provides a non-transitory computer readable storage medium having stored thereon a computer program which, when executed by a processor, is implemented to perform the artificial intelligence based motor pre-diagnosis methods provided above, the method comprising:
acquiring operation parameters affecting the state of a motor in the working machine;
inputting the operation parameters into a motor diagnosis model, and extracting a plurality of dimension characteristics from the operation parameters by the motor diagnosis model; the plurality of dimension features are used for representing the attributes of different dimensions of the operation parameters; the motor diagnosis model is constructed based on a feature decision tree, and the feature decision tree is constructed based on a plurality of sample dimension features corresponding to sample operation parameters and sample fault labels; the sample fault tag is used for representing the state of the motor corresponding to the sample operation parameter; the motor state includes a blocked state and an unblocked state;
and traversing a decision path in a feature decision tree based on the dimension features to obtain a motor diagnosis result of the motor.
The apparatus embodiments described above are merely illustrative, wherein the elements illustrated as separate elements may or may not be physically separate, and the elements shown as elements may or may not be physical elements, may be located in one place, or may be distributed over a plurality of network elements. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment. Those of ordinary skill in the art will understand and implement the present invention without undue burden.
From the above description of the embodiments, it will be apparent to those skilled in the art that the embodiments may be implemented by means of software plus necessary general hardware platforms, or of course may be implemented by means of hardware. Based on this understanding, the foregoing technical solution may be embodied essentially or in a part contributing to the prior art in the form of a software product, which may be stored in a computer readable storage medium, such as ROM/RAM, a magnetic disk, an optical disk, etc., including several instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the method described in the respective embodiments or some parts of the embodiments.
Finally, it should be noted that: the above embodiments are only for illustrating the technical solution of the present application, and are not limiting thereof; although the present application has been described in detail with reference to the foregoing embodiments, it should be understood by those of ordinary skill in the art that: the technical scheme described in the foregoing embodiments can be modified or some technical features thereof can be replaced by equivalents; such modifications and substitutions do not depart from the spirit and scope of the corresponding technical solutions.
Claims (10)
1. An artificial intelligence-based motor pre-diagnosis method, comprising the steps of:
acquiring operation parameters affecting the state of a motor in the working machine;
inputting the operation parameters into a motor diagnosis model, and extracting a plurality of dimension characteristics from the operation parameters by the motor diagnosis model; the plurality of dimension features are used for representing the attributes of different dimensions of the operation parameters; the motor diagnosis model is constructed based on a feature decision tree, and the feature decision tree is constructed based on a plurality of sample dimension features corresponding to sample operation parameters and sample fault labels; the sample fault tag is used for representing the state of the motor corresponding to the sample operation parameter; the motor state includes a blocked state and an unblocked state;
and traversing a decision path in a feature decision tree based on the dimension features to obtain a motor diagnosis result of the motor.
2. The motor pre-diagnosis method based on artificial intelligence according to claim 1, wherein the step of traversing a decision path in a feature decision tree based on each dimension feature to obtain a motor diagnosis result of the motor comprises the steps of:
taking a root node of the feature decision tree as a first starting node, and selecting a path corresponding to the dimension feature value along a branch of the feature decision tree to obtain a child node of the root node;
Selecting a path of corresponding dimension characteristic values along branches of the characteristic decision tree by taking the child node as a second starting node until the child node reaches leaf nodes of the characteristic decision tree;
and determining the fault state corresponding to the leaf node as the motor diagnosis result.
3. The artificial intelligence based motor pre-diagnosis method according to claim 1, wherein the constructing step of the feature decision tree comprises:
dividing: traversing each characteristic attribute corresponding to a current sample set, and dividing the current sample set based on each characteristic attribute of the current sample set to obtain a current sample sub-set corresponding to each characteristic attribute; the current sample set comprises all sample dimension characteristics corresponding to the sample operation parameters;
determining: determining information gain corresponding to the corresponding characteristic attribute based on the duty ratio of each sample fault label in each current sample subset;
the construction steps are as follows: taking a characteristic attribute corresponding to the maximum information gain as a dividing attribute of a current node of the characteristic decision tree, taking a current sample subset corresponding to the dividing attribute as the current sample set, and deleting sample dimension characteristics corresponding to the dividing attribute from the current sample set;
And returning to sequentially and repeatedly executing the dividing step, the determining step and the constructing step until no sample dimension characteristic exists in the current sample set.
4. The motor pre-diagnosis method based on artificial intelligence according to claim 3, wherein the determining the information gain corresponding to the corresponding feature attribute based on the duty ratio of the sample dimension feature corresponding to each sample fault label in each current sample subset comprises:
determining a first information entropy based on the duty ratio of various sample fault labels in each current sample subset; the first information entropy representation divides the current sample set according to the characteristic attribute to obtain an average value of data purity of each current sample subset; the data purity of each current sample subset is determined based on the proportion of each sample fault label in each current sample subset;
determining a second information entropy based on the duty ratio of various sample fault labels in the current sample set; the second information entropy represents the data purity of the current sample set based on the proportion of various sample fault labels in the current sample set;
and determining the information gain corresponding to the corresponding characteristic attribute based on the first information entropy and the second information entropy.
5. The artificial intelligence based motor pre-diagnosis method according to claim 3, further comprising, after constructing the feature decision tree:
acquiring a plurality of test dimension characteristics corresponding to test operation parameters and test fault labels corresponding to the test operation parameters; the test operation parameters represent operation parameters for testing the precision of the characteristic decision tree; the test dimension features are used for representing the attributes of different dimensions of the operation parameters;
applying each test dimension characteristic based on the characteristic decision tree, and determining a test fault result;
determining the precision of the feature decision tree based on the test fault result and the test fault label;
and under the condition that the precision of the feature decision tree is smaller than a threshold value, a plurality of sample dimension features and sample fault labels corresponding to the sample operation parameters are obtained in an increment mode, and the feature decision tree is updated based on the plurality of sample dimension features and the sample fault labels obtained in the increment mode.
6. The artificial intelligence based motor pre-diagnostic method of claim 5, wherein the determining the accuracy of the feature decision tree based on the test fault results and the test fault labels comprises:
Determining a real example, a false positive example, a true counterexample and a false counterexample based on the test fault result and the test fault label; the real case characterization correctly predicts the fault sample as a fault sample; the false positive token incorrectly predicts a non-faulty sample as a faulty sample; the true counterexample characterization correctly predicts a non-faulty sample as a non-faulty sample; the false counter characterization incorrectly predicts a failed sample as a non-failed sample;
determining a recall ratio and an precision ratio based on the true example, the false positive example, the true negative example and the false negative example; the recall ratio characterizes the recognition capability of the feature decision tree to a fault sample; the precision represents the accuracy of the feature decision tree in a sample predicted to be a fault;
and determining the precision of the feature decision tree based on the recall ratio and the precision ratio.
7. The artificial intelligence based motor pre-diagnosis method according to any one of claims 1 to 6, further comprising, prior to constructing the feature decision tree based on a plurality of sample dimension features and sample fault labels corresponding to sample operating parameters:
performing data processing on the sample operation parameters to obtain sample operation parameters after data processing;
Extracting a plurality of initial dimension features from the sample operation parameters after data processing; the plurality of initial dimensional features includes time domain dimensional features and frequency domain dimensional features;
and carrying out feature selection on each initial dimension feature to obtain the plurality of sample dimension features.
8. An artificial intelligence based motor pre-diagnosis device, comprising:
an acquisition unit for acquiring an operation parameter affecting a state of a motor in the work machine;
an extraction unit for inputting the operation parameters to a motor diagnosis model, and extracting a plurality of dimension features from the operation parameters by the motor diagnosis model; the plurality of dimension features are used for representing the attributes of different dimensions of the operation parameters; the motor diagnosis model is constructed based on a feature decision tree, and the feature decision tree is constructed based on a plurality of sample dimension features corresponding to sample operation parameters and sample fault labels; the sample fault tag is used for representing the state of the motor corresponding to the sample operation parameter; the motor state includes a blocked state and an unblocked state;
and the diagnosis unit is used for traversing a decision path in the feature decision tree based on the dimension features to obtain a motor diagnosis result of the motor.
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 implements the artificial intelligence based motor pre-diagnostic method of any one of claims 1 to 7 when the program is executed.
10. A non-transitory computer readable storage medium, having stored thereon a computer program which, when executed by a processor, implements the artificial intelligence based motor pre-diagnosis method of any of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410011112.0A CN117520964B (en) | 2024-01-04 | 2024-01-04 | Motor pre-diagnosis method based on artificial intelligence |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202410011112.0A CN117520964B (en) | 2024-01-04 | 2024-01-04 | Motor pre-diagnosis method based on artificial intelligence |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117520964A true CN117520964A (en) | 2024-02-06 |
| CN117520964B CN117520964B (en) | 2024-04-02 |
Family
ID=89745993
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202410011112.0A Active CN117520964B (en) | 2024-01-04 | 2024-01-04 | Motor pre-diagnosis method based on artificial intelligence |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117520964B (en) |
Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105590146A (en) * | 2016-02-29 | 2016-05-18 | 上海带来科技有限公司 | Power plant device intelligent prediction overhaul method and power plant device intelligent prediction overhaul system based on big data |
| CN108664010A (en) * | 2018-05-07 | 2018-10-16 | 广东省电信规划设计院有限公司 | Generating set fault data prediction technique, device and computer equipment |
| CN109492833A (en) * | 2018-12-25 | 2019-03-19 | 洛阳中科协同科技有限公司 | A kind of bearing ring quality of production prediction technique based on decision Tree algorithms |
| CN110188823A (en) * | 2019-05-30 | 2019-08-30 | 北京上格云技术有限公司 | A kind of Fault Tree Diagnosis Decision method and computer-readable medium |
| US20190302710A1 (en) * | 2018-03-30 | 2019-10-03 | General Electric Company | System and method for mechanical transmission control |
| CN110750655A (en) * | 2019-10-29 | 2020-02-04 | 西安科技大学 | Knowledge base optimization method of intelligent IETM fault maintenance auxiliary system |
| CN112855458A (en) * | 2019-11-26 | 2021-05-28 | 中车株洲电力机车研究所有限公司 | Anemometer fault diagnosis method, system and medium for wind generating set |
| CN115270946A (en) * | 2022-07-13 | 2022-11-01 | 上海高仙自动化科技发展有限公司 | Method and device for determining state of air suction equipment, electronic equipment and storage medium |
| CN116930750A (en) * | 2023-06-26 | 2023-10-24 | 青岛鹏海软件有限公司 | Motor monitoring and diagnosing method and system based on decision tree model and electronic equipment |
| CN117048524A (en) * | 2023-07-14 | 2023-11-14 | 深蓝汽车科技有限公司 | Method and device for detecting vehicle faults, vehicle and storage medium |
-
2024
- 2024-01-04 CN CN202410011112.0A patent/CN117520964B/en active Active
Patent Citations (10)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN105590146A (en) * | 2016-02-29 | 2016-05-18 | 上海带来科技有限公司 | Power plant device intelligent prediction overhaul method and power plant device intelligent prediction overhaul system based on big data |
| US20190302710A1 (en) * | 2018-03-30 | 2019-10-03 | General Electric Company | System and method for mechanical transmission control |
| CN108664010A (en) * | 2018-05-07 | 2018-10-16 | 广东省电信规划设计院有限公司 | Generating set fault data prediction technique, device and computer equipment |
| CN109492833A (en) * | 2018-12-25 | 2019-03-19 | 洛阳中科协同科技有限公司 | A kind of bearing ring quality of production prediction technique based on decision Tree algorithms |
| CN110188823A (en) * | 2019-05-30 | 2019-08-30 | 北京上格云技术有限公司 | A kind of Fault Tree Diagnosis Decision method and computer-readable medium |
| CN110750655A (en) * | 2019-10-29 | 2020-02-04 | 西安科技大学 | Knowledge base optimization method of intelligent IETM fault maintenance auxiliary system |
| CN112855458A (en) * | 2019-11-26 | 2021-05-28 | 中车株洲电力机车研究所有限公司 | Anemometer fault diagnosis method, system and medium for wind generating set |
| CN115270946A (en) * | 2022-07-13 | 2022-11-01 | 上海高仙自动化科技发展有限公司 | Method and device for determining state of air suction equipment, electronic equipment and storage medium |
| CN116930750A (en) * | 2023-06-26 | 2023-10-24 | 青岛鹏海软件有限公司 | Motor monitoring and diagnosing method and system based on decision tree model and electronic equipment |
| CN117048524A (en) * | 2023-07-14 | 2023-11-14 | 深蓝汽车科技有限公司 | Method and device for detecting vehicle faults, vehicle and storage medium |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117520964B (en) | 2024-04-02 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US8988236B2 (en) | System and method for failure prediction for rod pump artificial lift systems | |
| US20170097863A1 (en) | Detection method and information processing device | |
| CN105930963B (en) | Electromechanical system equipment health assessment method | |
| CN111597708A (en) | Method, system, readable medium and electronic device for compressor quality early warning service | |
| EP2916260A1 (en) | Time series analytics | |
| CN111538311B (en) | Flexible multi-state self-adaptive early warning method and device for mechanical equipment based on data mining | |
| CN112000081B (en) | Fault monitoring method and system based on multi-block information extraction and Mahalanobis distance | |
| CN111881594B (en) | Non-stationary signal state monitoring method and system for nuclear power equipment | |
| KR20210017651A (en) | Method for Fault Detection and Fault Diagnosis in Semiconductor Manufacturing Process | |
| Ceschini et al. | A Comprehensive Approach for Detection, Classification and Integrated Diagnostics of Gas Turbine Sensors (DCIDS) | |
| CN115935286A (en) | Abnormal point detection method, device and terminal for railway bearing state monitoring data | |
| CN114461534A (en) | Software performance testing method and system, electronic equipment and readable storage medium | |
| CN107122907B (en) | Method for analyzing symbolized quality characteristics of mechanical and electrical products and tracing fault reasons | |
| CN115392782A (en) | Method and system for monitoring and diagnosing health state of process system of nuclear power plant | |
| CN111176226A (en) | Automatic analysis method for alarm threshold of equipment characteristic parameter based on operation condition | |
| CN117633504B (en) | Optical fiber sensing evaluation method and device for state of oil immersed transformer | |
| CN104483958A (en) | Adaptive data driving fault diagnosis method and device in complex refining process | |
| CN117520964B (en) | Motor pre-diagnosis method based on artificial intelligence | |
| Mishra et al. | Hybrid models for rotating machinery diagnosis and prognosis: estimation of remaining useful life | |
| CN117312972A (en) | Method for identifying health state of scraper conveyor speed reducer | |
| CN117591860A (en) | Data anomaly detection method and device | |
| CN114674511B (en) | Bridge modal anomaly early warning method for eliminating time-varying environmental factor influence | |
| CN114112390B (en) | Nonlinear complex system early fault diagnosis method | |
| Chang et al. | Random forest-based multi-faults classification modeling and analysis for intelligent centrifugal pump system | |
| Ceschini et al. | Optimization of Statistical Methodologies for Anomaly Detection in Gas Turbine Dynamic Time Series |
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 |