CN115062553B - Water pump working condition degradation detection method, equipment and medium based on multi-model fusion - Google Patents
Water pump working condition degradation detection method, equipment and medium based on multi-model fusion Download PDFInfo
- Publication number
- CN115062553B CN115062553B CN202210983804.2A CN202210983804A CN115062553B CN 115062553 B CN115062553 B CN 115062553B CN 202210983804 A CN202210983804 A CN 202210983804A CN 115062553 B CN115062553 B CN 115062553B
- Authority
- CN
- China
- Prior art keywords
- data
- working condition
- detection
- water pump
- model
- 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.)
- Active
Links
- 238000001514 detection method Methods 0.000 title claims abstract description 189
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 title claims abstract description 109
- 238000006731 degradation reaction Methods 0.000 title claims abstract description 94
- 230000004927 fusion Effects 0.000 title claims abstract description 66
- 230000015556 catabolic process Effects 0.000 title claims abstract description 61
- 235000019580 granularity Nutrition 0.000 claims abstract description 73
- 238000000034 method Methods 0.000 claims abstract description 63
- 238000012549 training Methods 0.000 claims abstract description 42
- 238000013528 artificial neural network Methods 0.000 claims abstract description 30
- 238000012545 processing Methods 0.000 claims abstract description 20
- 238000004458 analytical method Methods 0.000 claims abstract description 18
- 230000008569 process Effects 0.000 claims description 31
- 230000007613 environmental effect Effects 0.000 claims description 13
- 238000004140 cleaning Methods 0.000 claims description 11
- 238000003860 storage Methods 0.000 claims description 11
- 238000013527 convolutional neural network Methods 0.000 claims description 10
- 239000011159 matrix material Substances 0.000 claims description 8
- 238000013507 mapping Methods 0.000 claims description 4
- 230000011218 segmentation Effects 0.000 claims description 4
- 238000012360 testing method Methods 0.000 claims description 3
- 238000010438 heat treatment Methods 0.000 claims description 2
- 238000012216 screening Methods 0.000 claims description 2
- 238000002372 labelling Methods 0.000 claims 2
- 238000004364 calculation method Methods 0.000 abstract description 5
- 238000007781 pre-processing Methods 0.000 abstract description 4
- 238000010586 diagram Methods 0.000 description 11
- 238000004590 computer program Methods 0.000 description 7
- 230000009286 beneficial effect Effects 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 230000006870 function Effects 0.000 description 4
- 230000005540 biological transmission Effects 0.000 description 3
- 238000007405 data analysis Methods 0.000 description 3
- 238000000605 extraction Methods 0.000 description 3
- 238000011897 real-time detection Methods 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000003287 optical effect Effects 0.000 description 2
- 238000007500 overflow downdraw method Methods 0.000 description 2
- 238000012300 Sequence Analysis Methods 0.000 description 1
- 230000002159 abnormal effect Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000004422 calculation algorithm Methods 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000006243 chemical reaction Methods 0.000 description 1
- 239000003245 coal Substances 0.000 description 1
- 238000003745 diagnosis Methods 0.000 description 1
- 239000000284 extract Substances 0.000 description 1
- 230000006872 improvement Effects 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000012821 model calculation Methods 0.000 description 1
- 238000013021 overheating Methods 0.000 description 1
- 230000000750 progressive effect Effects 0.000 description 1
- 230000000644 propagated effect Effects 0.000 description 1
- 238000013215 result calculation Methods 0.000 description 1
- 230000000717 retained effect Effects 0.000 description 1
- 230000003068 static effect Effects 0.000 description 1
- 230000017105 transposition Effects 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
- G06F30/27—Design optimisation, verification or simulation using machine learning, e.g. artificial intelligence, neural networks, support vector machines [SVM] or training a model
-
- F—MECHANICAL ENGINEERING; LIGHTING; HEATING; WEAPONS; BLASTING
- F04—POSITIVE - DISPLACEMENT MACHINES FOR LIQUIDS; PUMPS FOR LIQUIDS OR ELASTIC FLUIDS
- F04B—POSITIVE-DISPLACEMENT MACHINES FOR LIQUIDS; PUMPS
- F04B51/00—Testing machines, pumps, or pumping installations
Landscapes
- Engineering & Computer Science (AREA)
- Evolutionary Computation (AREA)
- Theoretical Computer Science (AREA)
- General Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Medical Informatics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Software Systems (AREA)
- Artificial Intelligence (AREA)
- Computer Hardware Design (AREA)
- Geometry (AREA)
- General Physics & Mathematics (AREA)
- Mechanical Engineering (AREA)
- Measurement Of Mechanical Vibrations Or Ultrasonic Waves (AREA)
- Testing Of Devices, Machine Parts, Or Other Structures Thereof (AREA)
Abstract
The application discloses a method, equipment and a medium for detecting degradation of working conditions of a water pump based on multi-model fusion, and relates to the field of computer systems based on specific calculation models, wherein the method comprises the following steps: acquiring detection data required by water pump analysis, wherein the detection data comprises working condition data and vibration data of the water pump, and the working condition data and the vibration data have different time granularities; the working condition detection model is obtained based on fusion training of a plurality of submodels, and different submodels are used for processing input data with different time granularities; and outputting the detection data to obtain the degradation state of the water pump on the basis of not unifying the time granularity of the working condition data and the vibration data. The preprocessing of the detection data is reduced, and the data information loss or redundancy caused by uniform time granularity is avoided, so that more useful information can be reserved in the data. And the neural network layers are fused to obtain a uniform neural network for training and predicting, so that the method is suitable for input of different time granularities, and the result accuracy is improved.
Description
Technical Field
The application relates to the field of computer systems based on specific calculation models, in particular to a water pump working condition degradation detection method, equipment and medium based on multi-model fusion.
Background
The water pump can be used in a plurality of industrial scenes, for example, in a coal mine heat exchange system, the water pump can continuously and automatically operate, and at the moment, the real-time detection on the working condition degradation of the water pump has quite important significance.
In the prior art, a time sequence data analysis algorithm can be used for realizing degradation detection on the working condition of the water pump. However, the multi-source data of the water pump usually have different time granularities, for the situation, a water pump working condition degradation detection model based on time sequence data analysis has certain difficulty in the aspect of unified time granularity input, and at the present stage, the traditional water pump working condition degradation detection still uses an industrial mechanism and subjective judgment, so that the time delay performance is certain, and the accuracy is low.
Disclosure of Invention
In order to solve the above problem, the present application provides a water pump working condition degradation detection method based on multi-model fusion, including:
acquiring detection data required by water pump analysis, wherein the detection data comprises working condition data and vibration data of the water pump, and the working condition data and the vibration data have different time granularities;
determining a pre-constructed working condition detection model, wherein the working condition detection model is obtained based on fusion training of a plurality of submodels, and different submodels are used for processing input data with different time granularities;
and directly inputting the detection data into the working condition detection model so as to output and obtain the degradation state of the water pump on the basis of not unifying the time granularity of the working condition data and the vibration data.
In one example, before determining the pre-constructed condition detection model, the method further comprises:
respectively building corresponding submodels aiming at different detection data, wherein the submodels at least comprise a first submodel which is built based on a one-dimensional convolutional neural network architecture and aims at working condition data and a second submodel which is built based on a two-dimensional convolutional neural network architecture and aims at vibration data, and the size of input data of each submodel is different from the size of convolution kernels of each convolution layer of the submodel;
fusing the sub-models through a preset model fusion framework, and performing judgment output by using a two-dimensional convolution layer and a Softmax layer to construct a working condition detection model;
and marking the training sample set of each submodel, and determining the input sequence of the training sample set of each submodel according to the connection sequence of the conticatenate layer in the model fusion framework.
In one example, the training process of the operating condition detection model comprises the following steps:
in the forward propagation process, output characteristic matrixes of the first sub-model and the second sub-model are spliced through a fusion layer in the model fusion frame, and characteristics of the spliced output characteristic matrixes are extracted through a layer behind the fusion layer, so that decision making is performed on the output layer;
in the backward propagation process, the loss calculated in the training process is transmitted to each neural network layer, and the full-connection layer coefficient and the matrix matched with the neural network layer by dividing the loss are updated through the fusion layer, wherein the neural network layer comprises the first submodel and the second submodel.
In one example, the working condition data is one-dimensional data 1*n, n is determined based on a required dimension in the working condition data, the vibration data is multidimensional data p × q, p represents vibration data in different directions, and q is obtained according to a working condition data acquisition interval and a vibration data acquisition frequency;
the output of the first submodel and the second submodel are one-dimensional data 1*m, and m is smaller than n.
In one example, the n corresponding dimensions include: at least a plurality of current, voltage, flow, spindle power, front axle temperature, rear axle temperature;
the detection data are directly input into the working condition detection model, so that the degradation state of the water pump is obtained through output on the basis that the working condition data and the vibration data are not unified in time granularity, and the method specifically comprises the following steps:
acquiring environmental data of the water pump, and performing preliminary analysis on the working condition data according to the environmental data to select a plurality of specified dimensions from the dimensions, wherein the probability of the specified dimensions corresponding to a degradation state is higher than a preset probability, and the plurality of dimensions are not more than n and are higher than m;
taking the appointed dimensionality as a first subset, selecting for multiple times in the remaining dimensionalities except the appointed dimensionality, and taking part of dimensionality obtained by each selection as a second subset;
respectively combining the first subset with each second subset to obtain a plurality of combined dimensions;
according to each combination dimension, working condition data corresponding to the combination dimension are used as input data in detection data and input to the working condition detection model, and the sub-degradation state of the water pump corresponding to the combination dimension is output and obtained on the basis that the working condition data and the vibration data are not subjected to time granularity unification;
and integrating all the sub-degradation states to obtain the degradation state of the water pump.
In one example, the sub-degradation states include: at least one of bearing damage, insufficient flow, rotor imbalance, rotor part looseness, motor heating and insufficient water quantity;
integrating all the sub-degradation states to obtain the degradation state of the water pump, specifically comprising:
determining, for each of the sub-states of degradation that occurs, a number of occurrences of that sub-state of degradation;
if the occurrence frequency is higher than the preset frequency, taking the sub-degradation state as the degradation state of the water pump;
and if the occurrence frequency is lower than the preset frequency, determining that the dimensionality corresponding to the sub-degradation state belongs to the second subset according to a preset mapping relation, and taking the sub-degradation state as the degradation state of the water pump.
In one example, noise data is also included in the detection data;
after collecting the detection data required by the water pump analysis, the method further comprises the following steps:
screening the detection data according to the time granularity of each detection data to obtain specified detection data of which the time granularity does not accord with the characteristics of the noise data, and taking the detection data of which the time granularity accords with the characteristics of the noise data as input data for inputting the working condition detection model;
and performing data cleaning on the specified detection data.
In one example, after performing data cleansing on the specified detection data, the method further comprises:
fitting and generating a current detection data curve according to the detection data after data cleaning;
determining a designated detection data curve closest to the current detection data curve in a historical detection data curve;
and generating virtual detection data according to the screened designated detection data in the data cleaning process and corresponding values in the designated detection data curve, and adding the virtual detection data into the detection data so as to use the detection data added with the virtual data to input the working condition detection model.
On the other hand, this application has still provided a water pump operating mode degradation check out test set based on multi-model fuses, includes:
at least one processor; and the number of the first and second groups,
a memory communicatively coupled to the at least one processor; wherein,
the memory stores instructions executable by the at least one processor to enable the at least one processor to:
acquiring detection data required by water pump analysis, wherein the detection data comprises working condition data and vibration data of the water pump, and the working condition data and the vibration data have different time granularities;
determining a pre-constructed working condition detection model, wherein the working condition detection model is obtained based on fusion training of a plurality of submodels, and different submodels are used for processing input data with different time granularities;
and directly inputting the detection data into the working condition detection model so as to output and obtain the degradation state of the water pump on the basis of not unifying the time granularity of the working condition data and the vibration data.
In another aspect, the present application further provides a non-volatile computer storage medium storing computer-executable instructions configured to:
acquiring detection data required by water pump analysis, wherein the detection data comprises working condition data and vibration data of the water pump, and the working condition data and the vibration data have different time granularities;
determining a pre-constructed working condition detection model, wherein the working condition detection model is obtained based on fusion training of a plurality of submodels, and different submodels are used for processing input data with different time granularities;
and directly inputting the detection data into the working condition detection model so as to output and obtain the degradation state of the water pump on the basis of not unifying the time granularity of the working condition data and the vibration data.
The water pump working condition degradation detection method based on multi-model fusion can bring the following beneficial effects:
the time granularity of the working condition data and the vibration data in the detection data is not unified, and the preprocessing of the detection data is reduced, so that the data information loss or redundancy caused by the unified time granularity is avoided, more useful information can be kept for the data of the water pump with different time granularities, and the working condition degradation detection result of the water pump is more accurate.
According to the method, the water pump data are detected in real time through a multi-submodel fusion method, different characteristics of a plurality of models are trained, then the models are fused for unified judgment and unified training, and judgment output of different types of fault information of the water pump is obtained, so that the time granularity of input data is reserved, data preprocessing is reduced, the authenticity of the data is reserved, and the accuracy of the models is improved.
The working condition detection model is finally obtained by fusing the neural network layers, one-time input and one-time output are taken as a unified neural network for training, so that the characteristics of each neural network training can be perceived by other neural networks in the training process, the characteristics of a plurality of input data with different time granularities are influenced mutually, the information of the input data is fully utilized, and more accurate decision is obtained.
The working condition detection model considers the characteristics of different time granularity data, the network is improved, the frequency domain characteristics of the convolution layer extracted water pump motor vibration data are utilized to make more accurate decision, and the fusion layer uses the uniform dimensionality of the full-connection layer and is trained. One-time training, one-time operation, both accuracy and calculation speed.
Drawings
The accompanying drawings, which are included to provide a further understanding of the application and are incorporated in and constitute a part of this application, illustrate embodiment(s) of the application and together with the description serve to explain the application and not to limit the application. In the drawings:
FIG. 1 is a schematic flow chart of a water pump working condition degradation detection method based on multi-model fusion in the embodiment of the present application;
FIG. 2 is a schematic flow chart of a water pump working condition degradation detection method based on multi-model fusion in one scenario in the embodiment of the present application;
FIG. 3 is a schematic diagram illustrating training of a condition detection model according to an embodiment of the present disclosure;
FIG. 4 is a schematic input/output diagram of a condition detection model in the embodiment of the present application;
FIG. 5 is a schematic diagram of water pump operating condition degradation detection equipment based on multi-model fusion in the embodiment of the present application.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the technical solutions of the present application will be described in detail and completely with reference to the following specific embodiments of the present application and the accompanying drawings. It should be apparent that the described embodiments are only some of the embodiments of the present application, and not all of the embodiments. 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 application.
The technical solutions provided by the embodiments of the present application are described in detail below with reference to the accompanying drawings.
In order to solve the above-mentioned technical problems, the present application provides several conventional approaches for performing real-time detection on a real-time data stream by using a time sequence data analysis method, for example, two state data of an instantaneous current and a water pump shaft power at a corresponding moment in the operation process of a water pump are obtained by a sensor mounted on the water pump, and an abnormal point analysis detection test of an operation state is performed by using a time sequence analysis model. Or, the real-time operation data of the water pump is analyzed and diagnosed by utilizing the neural network, the input data are data with coarse time granularity such as the lift, the flow, the power, the frequency and the like of the operation of the water pump, but the surface data are difficult to accurately reflect which kind of fault the water pump belongs to, and have certain hysteresis.
The water pump fault diagnosis can also be carried out by using fine time granularity data such as water pump vibration data, and the power frequency information is judged by comprehensively judging the waveforms of time domain and frequency domain conversion of the data, so that the fault information of the water pump is obtained, for example, frequency data is input, and different types of characteristic fault judgment are output. However, such data is relatively single, and is easy to cause misjudgment, for example, a sudden rise of voltage causes large burrs to appear on vibration data, which affects judgment.
Based on the method, the data of different time granularities are used as model input, and the real-time detection of the operation of the water pump is carried out. A traditional processing mode is provided, through unifying time granularity, input data are preprocessed, in the unification process, a part of fine time granularity data are abandoned, or coarse time granularity data are enriched, but the two methods can cause data distortion, the abandoned data can reduce data information to enable a model to be under-fitted, the enriched data can cause data redundancy to enable the model to be over-fitted, the model accuracy rate is difficult to improve, in addition, the processing mode for different types of data processed by the same model and also not in accordance with industrial data is provided, for example, common time sequence data and vibration data are different in signal processing mode.
Based on this, as shown in fig. 1, an embodiment of the present application provides a water pump operating condition degradation detection method based on multi-model fusion, including:
s101: the method comprises the steps of collecting detection data required by water pump analysis, wherein the detection data comprise working condition data and vibration data of the water pump, and the working condition data and the vibration data have different time granularities.
The working condition data may include data such as current, voltage, flow, spindle power, front axle temperature, rear axle temperature, or these multiple data are used as a dimension, and multi-dimensional data is obtained by combination and used as the working condition data. Generally, the condition data is second-level data, and the vibration data is millisecond-level data, which have different time granularities.
S102: determining a pre-constructed working condition detection model, wherein the working condition detection model is obtained based on fusion training of a plurality of submodels, and different submodels are used for processing input data with different time granularities.
If the time-granular reconstruction model of the input data is unified according to the above description, data distortion is easily caused. Therefore, the working condition detection model is obtained based on fusion training of a plurality of submodels, and different submodels are used for processing input data with different time granularities, so that the effect of judging various types of data is achieved.
Specifically, firstly, aiming at different detection data, corresponding submodels are respectively built, the submodels at least comprise a first submodel which is built based on a one-dimensional convolutional neural network architecture and aims at working condition data and a second submodel which is built based on a two-dimensional convolutional neural network architecture and aims at vibration data, and in the building process, the size of input data of each submodel is different from the size of convolution kernels of each convolution layer of the submodel.
Fusing a plurality of sub-models through a preset model fusion framework (for example, fusing a concatenate Layer as shown in fig. 2), judging by using a two-dimensional convolution Layer (Conv 2D) and a Softmax Layer, and outputting through an Output Layer (Output Layer) to construct a working condition detection model.
After the working condition detection model is built, the training sample sets of the submodels are labeled, the input sequence of the training sample sets of the submodels is determined according to the connection sequence of the concatenate layers in the model fusion framework, and the input sequence are adjusted to be consistent so that the working condition detection model can analyze data conveniently.
Further, as shown in fig. 2, the left side corresponds to a first sub-model of the operating condition data, which includes an Input Layer (Input Layer), the operating condition data is one-dimensional data 1 × n, n is determined based on the required dimension in the operating condition data, then the Input Layer is followed by a one-dimensional convolution Layer (conv 1D), and then a fully connected Layer (Dense) is connected, for example, when all the dimensions are selected, the operating condition data is one-dimensional data of 1*6. And the second sub-model corresponding to the vibration data on the right side comprises an Input Layer (Input Layer), the vibration data is multidimensional data p × q, wherein p represents the vibration data in different directions, q is obtained according to the working condition data acquisition interval and the vibration data acquisition frequency, two-dimensional convolution layers (Conv 2D) are arranged behind the Input Layer, and a transposition Layer (Transpose Layer) is connected behind the Input Layer. For example, p represents vibration data in three directions of x, y, and z, and at this time, the vibration data is multi-dimensional data 3*q. The outputs of the first submodel and the second submodel are all one-dimensional data 1 × m, m is smaller than n, for example, when n is 6, m can be 3, and at this time, the output is one-dimensional data of 1*3.
As shown in fig. 3, the working condition detection model in this document is finally obtained as a model, which is input once and Output once through fusion of neural Network layers, and is trained as a unified neural Network, so that a Loss (Loss) obtained in the training process can be propagated from an Output layer (Output) to each neural Network layer (including a first working layer Network1 of a first sub-model and a second working layer Network2 of a second sub-model) through a fusion layer (convergence) in a backward propagation process (that is, loss backoff propagation shown in fig. 3), and a coefficient (that is, update Parameters shown in fig. 3) is updated, so that characteristics of each neural Network training can be perceived by other neural networks in the training process, thereby achieving mutual influence of characteristics of input data of a plurality of different time granularities, fully utilizing information of the input data, and obtaining a more accurate decision. The fusion layer mainly performs two things, one output characteristic matrix of the two neural networks is spliced during forward transmission and serves as a full connection layer, the fusion output characteristic matrix further extracts characteristics, and then the characteristics are transmitted to the next layer; and the other is to update the coefficients of the full connection layer in backward propagation, and the segmentation loss (loss) is a matrix (such as a one-dimensional matrix) adapted to the neural network layer, and the loss is transmitted to the two neural networks according to the original sequence, so that the backward propagation is realized. And if the results are just fused, the loss of result calculation is not transmitted to each neural network layer, so that the accuracy of the result of the final training of the model is reduced, and the model is inferior to the working condition detection model mentioned in the text.
And when the model is transmitted in the forward direction, the output characteristic matrixes of the two submodels are spliced by the fusion layer in the model fusion frame, the subsequent layer of the fusion layer can carry out unified characteristic re-extraction (for example, the characteristic re-extraction is carried out by a third working layer Network 3) on the output characteristic matrixes of the two neural networks (more input data information is reserved compared with a single output value of result fusion), then the output layer carries out decision making, and the characteristics of input data of different time granularities of the water pump are fully utilized to carry out combined decision making. The result fusion input in the traditional scheme is a number or a judgment output value of a plurality of independent neural networks, the information quantity can be greatly reduced, and unified feature extraction is not carried out on the fused output, so that the result accuracy is reduced.
The working condition detection model considers the characteristics of different time granularity data, the network is improved, the frequency domain characteristics of the convolution layer extracted water pump motor vibration data are utilized to make more accurate decision, and the fusion layer uses the uniform dimensionality of the full-connection layer and is trained. One-time training and one-time operation are realized, and both the accuracy and the calculation speed are taken into consideration.
S103: and directly inputting the detection data into the working condition detection model so as to output and obtain the degradation state of the water pump on the basis of not unifying the time granularity of the working condition data and the vibration data.
As shown in fig. 4, when data is input, corresponding vibration data batch and operating condition data batch are input for the vibration data and the operating condition data, and enter a fusion model (that is, the operating condition detection model herein may also be referred to as a fusion model because it is obtained by fusion), so as to output the operating condition degradation type determination.
The time granularity unification is not carried out on the working condition data and the vibration data in the detection data, and the pretreatment on the detection data is reduced, so that the data information loss or redundancy caused by the unified time granularity is avoided, more useful information can be kept on the data of the water pump in different time granularities, and the detection result of the working condition degradation of the water pump is more accurate.
According to the method, the water pump data are detected in real time through a multi-submodel fusion method, different characteristics of a plurality of models are trained, then the models are fused and uniformly judged, uniform training is carried out, and judgment output of different types of fault information of the water pump is obtained, so that the time granularity of input data is reserved, data preprocessing is reduced, the authenticity of the data is reserved, and the accuracy of the models is improved.
In an embodiment, before the operating condition detection model is used for detection, environmental data (for example, environmental temperature data) where the water pump is located may be obtained first, and then preliminary analysis is performed on the operating condition data according to the environmental data, so as to select a plurality of specified dimensions in the dimensions, where the probability that the specified dimensions correspond to the degraded state is higher than a preset probability, and of course, the number of the specified dimensions is not more than n and is higher than m. For example, when the environmental temperature data is too high, the influence on the two dimensions of the front axle temperature and the rear axle temperature is the largest, and the probability of the corresponding degradation state is higher than the preset probability.
And taking the designated dimension as a first subset, selecting for multiple times in the remaining dimensions except the designated dimension, and taking the partial dimension obtained by each selection as a second subset. For example, the front axle temperature and the rear axle temperature are used as a first subset, the current and the voltage are selected for the first time to be used as a second subset, and the flow rate and the main shaft power are selected for the second time to be used as a second subset.
And respectively combining the first subset with each second subset to obtain a plurality of combined dimensions. For example, after combining the first subset and the second subset mentioned above, two combination dimensions are obtained, which are: front axle temperature, rear axle temperature, current, voltage; and front axle temperature, rear axle temperature, flow, spindle power.
According to each combination dimension, working condition data corresponding to the combination dimension are used as input data in detection data and input to a working condition detection model, and the sub-degradation state of the water pump corresponding to the combination dimension is output and obtained on the basis that the working condition data and the vibration data are not unified in time granularity; and all the sub-degradation states are integrated to obtain the degradation state of the water pump.
For the above two combined dimensions, they are input to the condition detection model, and the resulting output is taken as the sub-degradation state. Generally speaking, the first subset obtained by analyzing the environmental data is combined into a relatively important dimension, and is analyzed for multiple times, and the second subset is lower in importance, so that the times of analyzing the first subset are reduced, the performance of the important first subset in different dimension combinations can be considered more while a small amount of calculation is increased, and more accurate output is obtained by analyzing.
Further, the sub-degradation states include: bearing damage (for example, bearing damage caused by overheating of the bearing), insufficient flow (for example, insufficient flow caused by too low rotating speed), rotor imbalance, damage caused by severe vibration caused by loosening of parts of the rotor part (for example, damage caused by severe vibration caused by loosening of parts of the rotor part), motor heat (for example, motor heat caused by too high voltage), and insufficient water (for example, insufficient water caused by too low current).
When sub-degradation states are integrated, the number of occurrences of the sub-degradation state (the number of occurrences in the output result corresponding to all combination dimensions) is determined for each sub-degradation state that occurs. If the occurrence frequency is higher than the preset frequency, the sub-degradation state is probably occurred, and the sub-degradation state is taken as the degradation state of the water pump. If the occurrence frequency is lower than the preset frequency, determining that the dimension corresponding to the sub-degradation state belongs to the second subset according to a preset mapping relation (in the mapping relation, which dimension has the largest influence on the sub-degradation state, for example, the sub-degradation state with damaged bearing corresponds to two dimensions of front shaft temperature and rear shaft temperature), and then explaining the reason that the occurrence frequency of the sub-degradation state is small. If the dimension corresponding to the sub-degradation state belongs to the first subset, the first subset appears in each combined dimension, and the number of times of the sub-degradation state corresponding to the first subset appears is small, so that the output result is likely to be misjudged, and the output result is not taken as the degradation state of the water pump.
In one embodiment, the detection data generally includes noise data (the noise data is not noise generated by data transmission, data duplication generated in the processing process, field missing data, and the like, but is noise data generated by field collection due to the influence of factors such as the operating state of the water pump, the ambient environment, and the like). Based on this, after the detection data is acquired, the detection data is screened according to the time granularity of each detection data to acquire specified detection data in which the time granularity does not conform to the characteristics of the noise data, and the specified detection data is subjected to data cleaning. Typically, during data cleansing, noisy data is filtered out. In the present application, however, noise data is not screened out, but detection data with time granularity conforming to the characteristics of the noise data is used as input data for inputting an operating condition detection model for subsequent operating condition degradation detection.
Therefore, in the data cleansing, the data cleansing is performed in the specified detection data which does not conform to the characteristics of the noise data, for example, the noise data is usually in the millisecond order, and the data in the second order is subjected to the data cleansing, and the noise in which the data is repeated, the data is lacking in fields, and the like is removed, thereby the noise data is retained.
Further, with respect to the data after the cleaning, the final judgment may be inaccurate due to data missing. Based on the data, a current detection data curve is generated by fitting according to the detection data after data cleaning. The abscissa of the curve is time, and the ordinate is a value corresponding to the detected data. In the historical detected data curve, a specified detected data curve closest to the current detected data curve is determined.
According to the screened specified detection data in the data cleaning process, generating virtual detection data (the value of the data corresponding to the abscissa can be directly used as the virtual detection data, or after certain processing is performed on the value, the value is used as the virtual detection data) in a specified detection data curve, and adding the virtual detection data into the detection data so as to input the detection data added with the virtual data into a working condition detection model, thereby ensuring the integrity of the detection data and increasing the accuracy of model calculation.
As shown in fig. 5, an embodiment of the present application provides a water pump operating condition degradation detection apparatus based on multi-model fusion, including:
at least one processor; and the number of the first and second groups,
a memory communicatively coupled to the at least one processor; wherein,
the memory stores instructions executable by the at least one processor to enable the at least one processor to:
acquiring detection data required by water pump analysis, wherein the detection data comprises working condition data and vibration data of the water pump, and the working condition data and the vibration data have different time granularities;
determining a pre-constructed working condition detection model, wherein the working condition detection model is obtained based on fusion training of a plurality of submodels, and different submodels are used for processing input data with different time granularities;
and directly inputting the detection data into the working condition detection model so as to output and obtain the degradation state of the water pump on the basis of not unifying the time granularity of the working condition data and the vibration data.
An embodiment of the present application further provides a non-volatile computer storage medium storing computer-executable instructions, where the computer-executable instructions are configured to:
acquiring detection data required by water pump analysis, wherein the detection data comprises working condition data and vibration data of the water pump, and the working condition data and the vibration data have different time granularities;
determining a pre-constructed working condition detection model, wherein the working condition detection model is obtained based on fusion training of a plurality of submodels, and different submodels are used for processing input data with different time granularities;
and directly inputting the detection data into the working condition detection model so as to output and obtain the degradation state of the water pump on the basis of not unifying the time granularity of the working condition data and the vibration data.
The embodiments in the present application are described in a progressive manner, and the same and similar parts among the embodiments can be referred to each other, and each embodiment focuses on the differences from the other embodiments. In particular, for the device and media embodiments, the description is relatively simple as it is substantially similar to the method embodiments, and reference may be made to some descriptions of the method embodiments for relevant points.
The device and the medium provided by the embodiment of the application correspond to the method one to one, so the device and the medium also have the similar beneficial technical effects as the corresponding method, and the beneficial technical effects of the method are explained in detail above, so the beneficial technical effects of the device and the medium are not repeated herein.
As will be appreciated by one skilled in the art, embodiments of the present application may be provided as a method, system, or computer program product. Accordingly, the present application may take the form of an entirely hardware embodiment, an entirely software embodiment or an embodiment combining software and hardware aspects. Furthermore, the present application may take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, and so forth) having computer-usable program code embodied therein.
The present application is described with reference to flowchart illustrations and/or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the application. It will be understood that each flow and/or block of the flow diagrams and/or block diagrams, and combinations of flows and/or blocks in the flow diagrams and/or block diagrams, can be implemented by computer program instructions. These computer program instructions may be provided to a processor of a general purpose computer, special purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, create means for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be stored in a computer-readable memory that can direct a computer or other programmable data processing apparatus to function in a particular manner, such that the instructions stored in the computer-readable memory produce an article of manufacture including instruction means which implement the function specified in the flowchart flow or flows and/or block diagram block or blocks.
These computer program instructions may also be loaded onto a computer or other programmable data processing apparatus to cause a series of operational steps to be performed on the computer or other programmable apparatus to produce a computer implemented process such that the instructions which execute on the computer or other programmable apparatus provide steps for implementing the functions specified in the flowchart flow or flows and/or block diagram block or blocks.
In a typical configuration, a computing device includes one or more processors (CPUs), input/output interfaces, network interfaces, and memory.
The memory may include forms of volatile memory in a computer readable medium, random Access Memory (RAM) and/or non-volatile memory, such as Read Only Memory (ROM) or flash memory (flash RAM). Memory is an example of a computer-readable medium.
Computer-readable media, including both non-transitory and non-transitory, removable and non-removable media, may implement information storage by any method or technology. The information may be computer readable instructions, data structures, modules of a program, or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static Random Access Memory (SRAM), dynamic Random Access Memory (DRAM), other types of Random Access Memory (RAM), read Only Memory (ROM), electrically Erasable Programmable Read Only Memory (EEPROM), flash memory or other memory technology, compact disc read only memory (CD-ROM), digital Versatile Discs (DVD) or other optical storage, magnetic cassettes, magnetic tape magnetic disk storage or other magnetic storage devices, or any other non-transmission medium that can be used to store information that can be accessed by a computing device. As defined herein, a computer readable medium does not include a transitory computer readable medium such as a modulated data signal and a carrier wave.
It should also be noted that the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or apparatus. Without further limitation, an element defined by the phrase "comprising a … …" does not exclude the presence of another identical element in a process, method, article, or apparatus that comprises the element.
The above description is only an example of the present application and is not intended to limit the present application. Various modifications and changes may occur to those skilled in the art. Any modification, equivalent replacement, improvement or the like made within the spirit and principle of the present application shall be included in the scope of the claims of the present application.
Claims (6)
1. A water pump working condition degradation detection method based on multi-model fusion is characterized by comprising the following steps:
acquiring detection data required by analysis of a water pump, wherein the detection data comprises working condition data and vibration data of the water pump, and the working condition data and the vibration data have different time granularities;
determining a pre-constructed working condition detection model, wherein the working condition detection model is obtained based on fusion training of a plurality of submodels, and different submodels are used for processing input data with different time granularities;
directly inputting the detection data into the working condition detection model so as to output and obtain the degradation state of the water pump on the basis of not unifying the time granularity of the working condition data and the vibration data;
before determining the pre-constructed operating condition detection model, the method further comprises the following steps:
respectively building corresponding submodels aiming at different detection data, wherein the submodels at least comprise a first submodel which is built based on a one-dimensional convolutional neural network architecture and aims at working condition data and a second submodel which is built based on a two-dimensional convolutional neural network architecture and aims at vibration data, and the size of input data of each submodel is different from the size of convolution kernels of each convolution layer of the submodel;
fusing the plurality of sub-models through a preset model fusion framework, and performing judgment output by using a two-dimensional convolution layer and a Softmax layer to construct a working condition detection model;
labeling the training sample set of each submodel, and determining the input sequence of the training sample set of each submodel according to the connection sequence of the concatenate layer in the model fusion frame;
the training process of the working condition detection model comprises the following steps:
in the forward propagation process, output characteristic matrixes of the first sub-model and the second sub-model are spliced through a fusion layer in the model fusion frame, and characteristics of the spliced output characteristic matrixes are extracted through a layer behind the fusion layer, so that decision making is performed on the output layer;
in the backward propagation process, the loss calculated in the training process is transmitted to each neural network layer, and the full-connection layer coefficient and the segmentation loss are updated to be a matrix matched with the neural network layer through the fusion layer, wherein the neural network layer comprises the first submodel and the second submodel;
the working condition data is one-dimensional data 1*n, the n is determined based on the needed dimensionality in the working condition data, the vibration data is multidimensional data p x q, the p represents vibration data in different directions, and the q is obtained according to a working condition data acquisition interval and vibration data acquisition frequency; wherein, p represents vibration data in three directions of x, y and z, and the vibration data is multi-dimensional data of 3*q;
the output of the first submodel and the output of the second submodel are both one-dimensional data 1*m, and m is smaller than n;
the dimension corresponding to n comprises: at least a plurality of current, voltage, flow, spindle power, front axle temperature, rear axle temperature;
the detection data are directly input into the working condition detection model, so that the degradation state of the water pump is obtained through output on the basis that the working condition data and the vibration data are not unified in time granularity, and the method specifically comprises the following steps:
acquiring environmental data of the water pump, and performing preliminary analysis on the working condition data according to the environmental data to select a plurality of specified dimensions from the dimensions, wherein the probability of the specified dimensions corresponding to a degradation state is higher than a preset probability, and the plurality of dimensions are not more than n and are higher than m;
taking the appointed dimensionality as a first subset, selecting for multiple times in the remaining dimensionalities except the appointed dimensionality, and taking part of dimensionality obtained by each selection as a second subset;
respectively combining the first subset with each second subset to obtain a plurality of combined dimensions;
according to each combination dimension, working condition data corresponding to the combination dimension are used as input data in detection data and input to the working condition detection model, and the sub-degradation state of the water pump corresponding to the combination dimension is output and obtained on the basis that the working condition data and the vibration data are not subjected to time granularity unification;
and integrating all the sub-degradation states to obtain the degradation state of the water pump.
2. The water pump working condition degradation detection method based on multi-model fusion of claim 1, wherein the sub-degradation states comprise: at least one of bearing damage, insufficient flow, rotor imbalance, rotor part looseness, motor heating and insufficient water quantity;
integrating all the sub-degradation states to obtain the degradation state of the water pump, specifically comprising:
determining, for each of the sub-states of degradation that occur, a number of occurrences of that sub-state of degradation;
if the occurrence frequency is higher than the preset frequency, taking the sub-degradation state as the degradation state of the water pump;
and if the occurrence frequency is lower than the preset frequency, determining that the dimensionality corresponding to the sub-degradation state belongs to the second subset according to a preset mapping relation, and taking the sub-degradation state as the degradation state of the water pump.
3. The water pump working condition degradation detection method based on multi-model fusion of claim 1, wherein the detection data further comprises noise data;
after collecting the detection data required by the water pump analysis, the method further comprises the following steps:
screening the detection data according to the time granularity of each detection data to obtain specified detection data of which the time granularity does not accord with the characteristics of the noise data, and taking the detection data of which the time granularity accords with the characteristics of the noise data as input data for inputting the working condition detection model;
and performing data cleaning on the specified detection data.
4. The water pump working condition degradation detection method based on multi-model fusion as claimed in claim 3, wherein after data cleaning is performed on the specified detection data, the method further comprises:
fitting and generating a current detection data curve according to the detection data after data cleaning;
determining a designated detection data curve closest to the current detection data curve in a historical detection data curve;
and generating virtual detection data according to the screened designated detection data in the data cleaning process and corresponding values in the designated detection data curve, and adding the virtual detection data into the detection data so as to input the detection data added with the virtual detection data into the working condition detection model.
5. The utility model provides a water pump operating mode degradation check out test set based on multi-model fuses which characterized in that includes:
at least one processor; and the number of the first and second groups,
a memory communicatively coupled to the at least one processor; wherein,
the memory stores instructions executable by the at least one processor to enable the at least one processor to:
acquiring detection data required by analysis of a water pump, wherein the detection data comprises working condition data and vibration data of the water pump, and the working condition data and the vibration data have different time granularities;
determining a pre-constructed working condition detection model, wherein the working condition detection model is obtained based on fusion training of a plurality of submodels, and different submodels are used for processing input data with different time granularities;
directly inputting the detection data into the working condition detection model so as to output and obtain the degradation state of the water pump on the basis of not unifying the time granularity of the working condition data and the vibration data;
before determining the pre-constructed operating condition detection model, the method further comprises the following steps:
respectively building corresponding submodels aiming at different detection data, wherein the submodels at least comprise a first submodel which is built based on a one-dimensional convolutional neural network architecture and aims at working condition data and a second submodel which is built based on a two-dimensional convolutional neural network architecture and aims at vibration data, and the size of input data of each submodel is different from the size of convolution kernels of each convolution layer of the submodel;
fusing the plurality of sub-models through a preset model fusion framework, and performing judgment output by using a two-dimensional convolution layer and a Softmax layer to construct a working condition detection model;
marking the training sample set of each submodel, and determining the input sequence of the training sample set of each submodel according to the connection sequence of the conticatenate layer in the model fusion framework;
the training process of the working condition detection model comprises the following steps:
in the forward propagation process, output characteristic matrixes of the first sub-model and the second sub-model are spliced through a fusion layer in the model fusion frame, and characteristics of the spliced output characteristic matrixes are extracted through a layer behind the fusion layer, so that decision making is performed on the output layer;
in the backward propagation process, the loss calculated in the training process is transmitted to each neural network layer, and the full-connection layer coefficient and the segmentation loss are updated to be a matrix matched with the neural network layer through the fusion layer, wherein the neural network layer comprises the first submodel and the second submodel;
the working condition data is one-dimensional data 1*n, the n is determined based on the needed dimensionality in the working condition data, the vibration data is multidimensional data p x q, the p represents vibration data in different directions, and the q is obtained according to a working condition data acquisition interval and vibration data acquisition frequency; wherein, p represents vibration data in x, y and z directions, and the vibration data is multi-dimensional data of 3*q;
the output of the first submodel and the second submodel are one-dimensional data 1*m, and m is smaller than n;
the dimension corresponding to n comprises: at least a plurality of current, voltage, flow, spindle power, front axle temperature, rear axle temperature;
the detection data are directly input into the working condition detection model, so that the degradation state of the water pump is obtained through output on the basis that the working condition data and the vibration data are not unified in time granularity, and the method specifically comprises the following steps:
acquiring environmental data of the water pump, and performing preliminary analysis on the working condition data according to the environmental data to select a plurality of specified dimensions from the dimensions, wherein the probability of the specified dimensions corresponding to a degradation state is higher than a preset probability, and the plurality of dimensions are not more than n and are higher than m;
taking the specified dimensionality as a first subset, selecting for multiple times in the remaining dimensionalities except the specified dimensionality, and taking part of dimensionality obtained by each selection as a second subset;
respectively combining the first subset with each second subset to obtain a plurality of combined dimensions;
according to each combination dimension, working condition data corresponding to the combination dimension are used as input data in detection data and input to the working condition detection model, and the sub-degradation state of the water pump corresponding to the combination dimension is output and obtained on the basis that the working condition data and the vibration data are not subjected to time granularity unification;
and integrating all the sub-degradation states to obtain the degradation state of the water pump.
6. A non-transitory computer storage medium storing computer-executable instructions, the computer-executable instructions configured to:
acquiring detection data required by water pump analysis, wherein the detection data comprises working condition data and vibration data of the water pump, and the working condition data and the vibration data have different time granularities;
determining a pre-constructed working condition detection model, wherein the working condition detection model is obtained based on fusion training of a plurality of submodels, and different submodels are used for processing input data with different time granularities;
directly inputting the detection data into the working condition detection model so as to output and obtain the degradation state of the water pump on the basis of not unifying the time granularity of the working condition data and the vibration data;
before determining the pre-constructed working condition detection model, the method further comprises the following steps:
respectively building corresponding submodels aiming at different detection data, wherein the submodels at least comprise a first submodel which is built based on a one-dimensional convolutional neural network architecture and aims at working condition data and a second submodel which is built based on a two-dimensional convolutional neural network architecture and aims at vibration data, and the size of input data of each submodel is different from the size of convolution kernels of each convolution layer of the submodel;
fusing the plurality of sub-models through a preset model fusion framework, and performing judgment output by using a two-dimensional convolution layer and a Softmax layer to construct a working condition detection model;
labeling the training sample set of each submodel, and determining the input sequence of the training sample set of each submodel according to the connection sequence of the concatenate layer in the model fusion frame;
the training process of the working condition detection model comprises the following steps:
in the forward propagation process, output characteristic matrixes of the first sub-model and the second sub-model are spliced through a fusion layer in the model fusion frame, and characteristics of the spliced output characteristic matrixes are extracted through a layer behind the fusion layer, so that decision making is performed on the output layer;
in the backward propagation process, the loss calculated in the training process is transmitted to each neural network layer, and the full-connection layer coefficient and the segmentation loss are updated to be a matrix matched with the neural network layer through the fusion layer, wherein the neural network layer comprises the first submodel and the second submodel;
the working condition data is one-dimensional data 1*n, the n is determined based on the needed dimensionality in the working condition data, the vibration data is multidimensional data p x q, the p represents vibration data in different directions, and the q is obtained according to a working condition data acquisition interval and vibration data acquisition frequency; wherein, p represents vibration data in three directions of x, y and z, and the vibration data is multi-dimensional data of 3*q;
the output of the first submodel and the output of the second submodel are both one-dimensional data 1*m, and m is smaller than n;
the corresponding dimension of n includes: at least a plurality of current, voltage, flow, spindle power, front axle temperature, rear axle temperature;
the detection data are directly input into the working condition detection model, so that the degradation state of the water pump is obtained through output on the basis that the working condition data and the vibration data are not unified in time granularity, and the method specifically comprises the following steps:
acquiring environmental data of the water pump, and performing preliminary analysis on the working condition data according to the environmental data to select a plurality of specified dimensions from the dimensions, wherein the probability of the specified dimensions corresponding to a degradation state is higher than a preset probability, and the plurality of dimensions are not more than n and are higher than m;
taking the specified dimensionality as a first subset, selecting for multiple times in the remaining dimensionalities except the specified dimensionality, and taking part of dimensionality obtained by each selection as a second subset;
respectively combining the first subset with each second subset to obtain a plurality of combined dimensions;
according to each combination dimension, working condition data corresponding to the combination dimension are used as input data in detection data and input to the working condition detection model, and the sub-degradation state of the water pump corresponding to the combination dimension is output and obtained on the basis that the working condition data and the vibration data are not subjected to time granularity unification;
and integrating all the sub-degradation states to obtain the degradation state of the water pump.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210983804.2A CN115062553B (en) | 2022-08-17 | 2022-08-17 | Water pump working condition degradation detection method, equipment and medium based on multi-model fusion |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202210983804.2A CN115062553B (en) | 2022-08-17 | 2022-08-17 | Water pump working condition degradation detection method, equipment and medium based on multi-model fusion |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN115062553A CN115062553A (en) | 2022-09-16 |
| CN115062553B true CN115062553B (en) | 2022-11-25 |
Family
ID=83207752
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202210983804.2A Active CN115062553B (en) | 2022-08-17 | 2022-08-17 | Water pump working condition degradation detection method, equipment and medium based on multi-model fusion |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN115062553B (en) |
Families Citing this family (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117076928A (en) * | 2023-08-25 | 2023-11-17 | 中交路桥科技有限公司 | Bridge health state monitoring method, device and system and electronic equipment |
| CN116878886B (en) * | 2023-09-05 | 2023-12-15 | 济宁鲁威液压科技股份有限公司 | Method for monitoring abrasion of water pump bearing of cooling tower |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108343599A (en) * | 2018-01-10 | 2018-07-31 | 中国水利水电科学研究院 | A kind of water pump assembly intelligent failure diagnosis method cascading forest based on more granularities |
| CN113673442A (en) * | 2021-08-24 | 2021-11-19 | 燕山大学 | Variable working condition fault detection method based on semi-supervised single classification network |
| EP3913453A1 (en) * | 2020-05-15 | 2021-11-24 | Deere & Company | Fault detection system and method for a vehicle |
| CN114154077A (en) * | 2021-10-21 | 2022-03-08 | 北京邮电大学 | Multi-dimensional fine-grained dynamic emotion analysis method and system |
| CN114357858A (en) * | 2021-12-06 | 2022-04-15 | 苏州方正璞华信息技术有限公司 | Equipment deterioration analysis method and system based on multi-task learning model |
| CN114548154A (en) * | 2022-01-21 | 2022-05-27 | 中国核电工程有限公司 | Intelligent diagnosis method and device for important service water pump |
Family Cites Families (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| US11378947B2 (en) * | 2020-07-02 | 2022-07-05 | AB Cognitive Systems Inc. | System and methods of failure prediction and prevention for rotating electrical machinery |
| CN114648075A (en) * | 2022-04-07 | 2022-06-21 | 深圳依时货拉拉科技有限公司 | Information processing method, information processing apparatus, storage medium, and electronic device |
-
2022
- 2022-08-17 CN CN202210983804.2A patent/CN115062553B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108343599A (en) * | 2018-01-10 | 2018-07-31 | 中国水利水电科学研究院 | A kind of water pump assembly intelligent failure diagnosis method cascading forest based on more granularities |
| EP3913453A1 (en) * | 2020-05-15 | 2021-11-24 | Deere & Company | Fault detection system and method for a vehicle |
| CN113673442A (en) * | 2021-08-24 | 2021-11-19 | 燕山大学 | Variable working condition fault detection method based on semi-supervised single classification network |
| CN114154077A (en) * | 2021-10-21 | 2022-03-08 | 北京邮电大学 | Multi-dimensional fine-grained dynamic emotion analysis method and system |
| CN114357858A (en) * | 2021-12-06 | 2022-04-15 | 苏州方正璞华信息技术有限公司 | Equipment deterioration analysis method and system based on multi-task learning model |
| CN114548154A (en) * | 2022-01-21 | 2022-05-27 | 中国核电工程有限公司 | Intelligent diagnosis method and device for important service water pump |
Non-Patent Citations (2)
| Title |
|---|
| Multi-kernel Learning based Autonomous Fault Diagnosis for Centrifugal Pumps;H. Feng 等;《2018 International Conference on Control, Automation and Information Sciences (ICCAIS)》;20181209;第548-553页 * |
| 水电机组振动劣化预警模型研究及应用;桂中华 等;《水利学报》;20180211;第49卷(第02期);第216-222页 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN115062553A (en) | 2022-09-16 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN115062553B (en) | Water pump working condition degradation detection method, equipment and medium based on multi-model fusion | |
| Wang et al. | Dual-attention generative adversarial networks for fault diagnosis under the class-imbalanced conditions | |
| CN107877262A (en) | A kind of numerical control machine tool wear monitoring method based on deep learning | |
| CN115800272B (en) | Power grid fault analysis method, system, terminal and medium based on topology identification | |
| CN112748359A (en) | Power distribution network ground fault identification method and system based on random forest | |
| CN115453356B (en) | Power equipment operation state monitoring and analyzing method, system, terminal and medium | |
| CN114548199A (en) | Multi-sensor data fusion method based on deep migration network | |
| CN107689059A (en) | The abnormal recognition methods of wind generating set pitch control and device | |
| CN116430817B (en) | Data acquisition processing method and system applied to photovoltaic module production system | |
| CN116085212B (en) | Method and system for monitoring running state of new energy wind turbine generator in real time | |
| CN116625686A (en) | On-line diagnosis method for bearing faults of aero-engine | |
| CN115389865A (en) | Power equipment fault diagnosis method and system based on cloud edge terminal cooperative architecture | |
| CN117877521A (en) | Data enhancement-based unsupervised mechanical sound abnormality identification method and system | |
| CN116820821A (en) | Disk failure detection method, apparatus, electronic device and computer readable storage medium | |
| CN116482460A (en) | Fault diagnosis method for power grid equipment and related equipment | |
| KR102466891B1 (en) | Apparatus and method for diagnosis of fault of pump turbine of pumped-storage hydroelectricity system | |
| US11144046B2 (en) | Fault signal recovery apparatus and method | |
| CN113311364B (en) | Permanent magnet synchronous motor inverter open-circuit fault diagnosis method based on multi-core SVM | |
| CN115795397A (en) | Gearbox temperature prediction method based on 1DCNN-LSTM and BiLSTM parallel network | |
| CN110826904B (en) | Data processing method and device for fan, processing equipment and readable storage medium | |
| CN114330569A (en) | Method and equipment for detecting fan unit component fault and storage medium | |
| CN106529805A (en) | Power generation system reliability evaluation method based on power generator importance | |
| WO2024212410A1 (en) | Battery cell anomaly determination method and system, processing device, and storage medium | |
| CN116658489B (en) | Hydraulic system fault diagnosis method and system based on digital twinning | |
| CN112459970B (en) | Wind turbine generator anomaly detection method using data counterstudy |
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 | ||
| TR01 | Transfer of patent right | ||
| TR01 | Transfer of patent right |
Effective date of registration: 20240628 Address after: Room 801, 8th Floor, S01 Building, No. 1036 Inspur Road, High tech Zone, Jinan City, Shandong Province, 250013 Patentee after: Inspur City Service Technology Co.,Ltd. Country or region after: China Address before: 250101 Inspur science and Technology Park, 1036 Inspur Road, hi tech Zone, Jinan City, Shandong Province Patentee before: Inspur Genersoft Co.,Ltd. Country or region before: China |