CN112882909A - Fuel cell system fault prediction method and device - Google Patents
Fuel cell system fault prediction method and device Download PDFInfo
- Publication number
- CN112882909A CN112882909A CN201911196734.0A CN201911196734A CN112882909A CN 112882909 A CN112882909 A CN 112882909A CN 201911196734 A CN201911196734 A CN 201911196734A CN 112882909 A CN112882909 A CN 112882909A
- Authority
- CN
- China
- Prior art keywords
- fault
- real
- data
- fuel cell
- cell system
- 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.)
- Pending
Links
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/30—Monitoring
- G06F11/34—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
- G06F11/3409—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment for performance assessment
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F11/00—Error detection; Error correction; Monitoring
- G06F11/30—Monitoring
- G06F11/34—Recording or statistical evaluation of computer activity, e.g. of down time, of input/output operation ; Recording or statistical evaluation of user activity, e.g. usability assessment
- G06F11/3466—Performance evaluation by tracing or monitoring
- G06F11/3476—Data logging
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Engineering & Computer Science (AREA)
- General Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- Computing Systems (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Biomedical Technology (AREA)
- Molecular Biology (AREA)
- Biophysics (AREA)
- Artificial Intelligence (AREA)
- Life Sciences & Earth Sciences (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Computer Hardware Design (AREA)
- Quality & Reliability (AREA)
- Fuel Cell (AREA)
Abstract
The application relates to a fuel cell system fault prediction method, a device, computer equipment and a storage medium. The method comprises the following steps: the method comprises the steps of obtaining real-time operation index data of a fuel cell system, preprocessing the real-time operation index data to obtain preprocessed real-time operation index data, inputting the preprocessed real-time operation index data into a fault index prediction model to obtain real-time fault characteristic indexes of the fuel cell system, and inputting the real-time fault characteristic indexes of the fuel cell system into a fault classification model to obtain real-time fault codes of the fuel cell system. The method can realize the accuracy, timeliness and convenience of the fault prediction of the fuel cell system.
Description
Technical Field
The present application relates to the field of intelligent technologies, and in particular, to a method and an apparatus for predicting a failure of a fuel cell system, a computer device, and a storage medium.
Background
A hydrogen fuel cell system is a complex set of nonlinear systems including a stack, an air supply system, a hydrogen gas supply system, a thermal management system, etc. The operation data of the hydrogen fuel cell system is monitored in real time in the use process, the fault state of the hydrogen fuel cell system is estimated according to the operation data, reasonable suggestions are given to the maintenance of key components in time, and the method is of great importance to the safe and stable operation of the system.
The failure prediction methods are mainly classified into three types, namely a prediction method based on a mechanism model, a prediction method based on a data model and a hybrid model prediction method. 1) The mechanism model-based prediction method comprises the following steps: aiming at the failure-based physical model of the electronic component, the input and output model of the system is based; 2) the prediction method based on probability statistics comprises the following steps: predicting the fault from the angle of statistical characteristics of past fault historical data; 3) prediction method based on data driving: and (3) performing system fault prediction by using the test and sensor historical data of each stage of component or system design, simulation, operation, maintenance and the like.
The fault prediction method in the prior art is difficult to analyze the interaction of each fault on the system under the condition of compound faults, analyze the complex characteristics of the data of the hydrogen fuel cell system and accurately predict the faults in real time.
Disclosure of Invention
In view of the above, it is necessary to provide a fuel cell system fault prediction method, apparatus, computer device and storage medium capable of improving accuracy, timeliness and convenience of fuel cell system fault prediction.
A fuel cell system failure prediction method, the method comprising:
acquiring real-time operation index data of a fuel cell system;
preprocessing the real-time operation index data to obtain preprocessed real-time operation index data;
inputting the preprocessed real-time operation index data into a fault index prediction model to obtain a real-time fault characteristic index of the fuel cell system;
and inputting the real-time fault characteristic indexes of the fuel cell system into a fault classification model to obtain real-time fault codes of the fuel cell system.
In one embodiment, the preprocessing the real-time operation index data to obtain the preprocessed real-time operation index data includes:
and sequentially filtering, complementing and standardizing the real-time operation index data to obtain the preprocessed real-time operation index data.
In one embodiment, the sequentially filtering, complementing and standardizing the real-time operation index data to obtain the preprocessed real-time operation index data includes:
acquiring historical operating data of a fuel cell system within preset time;
preprocessing the historical operating data to obtain preprocessed historical operating data;
determining a first target training sample according to the preprocessed historical operating data;
and training the LSTM network by adopting the first target training sample to determine a fault index prediction model.
In one embodiment, the determining a first target training sample according to the preprocessed historical operating data includes:
extracting fault characteristic data in the preprocessed historical operating data, and taking the fault characteristic data and the preprocessed historical operating data as a first initial training sample;
and sequentially filtering, complementing and standardizing the data in the first initial training sample to obtain a first target training sample.
In one embodiment, the training the LSTM network with the first target training sample, and determining the fault indicator prediction model includes:
obtaining an LSTM network and a first target training sample;
and inputting the first target training sample into an LSTM network, and training the LSTM network by using a gradient descent method to determine a fault index prediction model.
In one embodiment, the inputting the first target training sample into the LSTM network and training the LSTM network by using a gradient descent method to determine the fault indicator prediction model includes:
and recovering the fault index prediction model by using TensorFlow.
In one embodiment, the recovering the fault indicator prediction model by using TensorFlow comprises:
acquiring the fault characteristic data and a fault code corresponding to the fault characteristic data, and taking the fault characteristic data and the fault code corresponding to the fault characteristic data as a second initial training sample;
sequentially filtering, complementing and encoding the data in the second initial training sample to obtain a second target training sample;
acquiring a CNN network;
and training the CNN network by adopting the second target training sample to determine a fault classification model.
In one embodiment, the training of the CNN network with the second target training sample, after determining the fault classification model, includes:
and recovering the fault classification model by using TensorFlow.
In one embodiment, the recovering the fault classification model by using the TensorFlow comprises the following steps:
inputting the real-time fault characteristic index of the fuel cell system into the fault classification model, and outputting a one-hot coded probability value;
and acquiring an index value of the maximum probability value in the probability values, and acquiring the real-time fault code of the fuel cell system through inverse transformation of the one-hot code.
In one embodiment, the obtaining of the index value of the maximum probability value in the probability values after obtaining the real-time fault code of the fuel cell system through inverse transformation of one-hot coding comprises:
and carrying out persistence processing on the real-time fault code and reporting an alarm.
A fuel cell system failure prediction apparatus, the apparatus comprising:
the first acquisition module is used for acquiring real-time operation index data of the fuel cell system;
the first processing module is used for preprocessing the real-time operation index data to obtain preprocessed real-time operation index data;
the first determination module is used for inputting the preprocessed real-time operation index data into a fault index prediction model to obtain a real-time fault characteristic index of the fuel cell system;
and the second determination module is used for inputting the real-time fault characteristic indexes of the fuel cell system into a fault classification model to obtain real-time fault codes of the fuel cell system.
A computer device comprising a memory storing a computer program and a processor implementing the steps of the method as claimed in any one of the above when the computer program is executed.
A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the steps of the method according to any one of the preceding claims.
According to the method, the device, the computer equipment and the storage medium for predicting the fault of the fuel cell system, the real-time operation index data of the fuel cell system is obtained, then the real-time operation index data is preprocessed, the preprocessed real-time operation index data is input into a fault index prediction model to obtain the real-time fault characteristic index of the fuel cell system, and the real-time fault characteristic index of the fuel cell system is input into a fault classification model to obtain the real-time fault code of the fuel cell system. The method combines big data and a machine learning method to process the operation index data and the fault characteristic index of the fuel cell system so as to realize the accuracy, timeliness and convenience of the fault prediction of the fuel cell system.
Drawings
FIG. 1 is a diagram of an exemplary embodiment of a fuel cell system failure prediction method;
FIG. 2 is a schematic flow diagram of a method for predicting a fuel cell system failure according to one embodiment;
FIG. 3 is a graphical representation of the tan h function in one embodiment;
FIG. 4 is a graph illustrating a sigmod function in one embodiment;
fig. 5 is a block diagram showing the configuration of a fuel cell system failure prediction apparatus according to an embodiment;
FIG. 6 is a diagram illustrating an internal structure of a computer device according to an embodiment.
Detailed Description
In order to make the objects, technical solutions and advantages of the present application more apparent, the present application is described in further detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative of the present application and are not intended to limit the present application.
The fuel cell system fault prediction method provided by the application can be applied to the application environment shown in fig. 1. Wherein the fuel cell system 102 communicates with the server 104 via a network. The server 104 acquires real-time operation index data of the fuel cell system, and preprocesses the real-time operation index data to obtain preprocessed real-time operation index data. Inputting the preprocessed real-time operation index data into a fault index prediction model to obtain a real-time fault characteristic index of the fuel cell system; and inputting the real-time fault characteristic indexes of the fuel cell system into a fault classification model to obtain real-time fault codes of the fuel cell system. The server 104 may be implemented as a stand-alone server or a server cluster composed of a plurality of servers.
In one embodiment, as shown in fig. 2, a method for predicting a failure of a fuel cell system is provided, which is illustrated by applying the method to the server 104 in fig. 1, and includes the following steps:
step S1: acquiring real-time operation index data of a fuel cell system;
step S2: preprocessing the real-time operation index data to obtain preprocessed real-time operation index data;
step S3: inputting the preprocessed real-time operation index data into a fault index prediction model to obtain a real-time fault characteristic index of the fuel cell system;
step S4: and inputting the real-time fault characteristic indexes of the fuel cell system into a fault classification model to obtain real-time fault codes of the fuel cell system.
In steps S1-S4, the real-time operation index data of the fuel cell system may include parameters such as current, voltage, power or time, for example, the real-time operation index data is data in one period, i.e., current, voltage, power and time corresponding to each time point in one period. The real-time operation data adopted by the method can be current, voltage, power and time at a certain time point, and can also be current, voltage, power and time corresponding to each time point in a period. Because the collected real-time operation index data may have an abnormal value or a null value, the real-time operation index data needs to be preprocessed in advance to obtain an accurate real-time fault code.
The failure index prediction model may be a single model or a combination of models including a plurality of failure index prediction models. The fault index prediction model is a calculation model which inputs historical operation data in a period of time into the fault index prediction model, performs prediction analysis on future changes of the fault index prediction model, and further predicts possible data in a period of time in the future. There are many ways to implement the fault indicator prediction model, such as Long Short Term memory network (LSTM).
The fault classification model may be a single model or a combination of models including a plurality of fault classification submodels. The fault classification model may be implemented in many ways, and may be a machine learning-based classification model such as a Convolutional Neural Network (CNN), a naive bayes algorithm model, a support vector machine algorithm model, or the like.
According to the method for predicting the fault of the fuel cell system, real-time operation index data of the fuel cell system is obtained, then the real-time operation index data is preprocessed, the preprocessed real-time operation index data is input into a fault index prediction model, real-time fault characteristic indexes of the fuel cell system are obtained, the real-time fault characteristic indexes of the fuel cell system are input into a fault classification model, and real-time fault codes of the fuel cell system are obtained. The method combines big data and a machine learning method to process the operation index data and the fault characteristic index of the fuel cell system so as to realize the accuracy, timeliness and convenience of the fault prediction of the fuel cell system.
In one embodiment, the step S2 includes:
step S21: and sequentially filtering, complementing and standardizing the real-time operation index data to obtain the preprocessed real-time operation index data.
Specifically, the preprocessing of the real-time operation index data at least includes filling all Nan data to 0, where the Nan data is a null value. For example, when real-time operation index data is collected, since the network has no signal, a current value or a voltage value at a certain time point cannot be collected, and then the index that cannot be collected needs to be filled.
In one embodiment, the step S21 is followed by:
step S22: acquiring historical operating data of a fuel cell system within preset time;
step S23: preprocessing the historical operating data to obtain preprocessed historical operating data;
step S24: determining a first target training sample according to the preprocessed historical operating data;
step S25: and training the LSTM network by adopting the first target training sample to determine a fault index prediction model.
In steps S22-S25, the preset time refers to a continuous time period, i.e., historical operation data of a certain period is required to be acquired. The historical operating data may include one or more of the following indicators of the fuel cell system, such as a voltage value, a current value, or an operating time of the fuel cell system. For example, the historical operating data may be a historical feature vector constructed with "voltage value, current value, and operating time". It should be noted that the indicators in the historical operating data are of the same type as the indicators in the real-time operating indicator data.
Furthermore, the historical operating data is preprocessed, the integrity and the accuracy of the historical operating data are enhanced, and the reliability of the training sample can be improved, so that the accuracy of target data (namely real-time fault coding) to be acquired is improved.
In one embodiment, the TensorFlow is a symbolic mathematical system based on data flow programming, is widely applied to programming of various machine learning (machine learning) algorithms to realize that the TensorFlow has a multi-level structure, can be deployed in various servers, PC terminals and webpages and supports GPU and TPU high-performance numerical calculation, and is widely applied to product development and scientific research in various fields. The method comprises the steps that an LSTM network is established based on TensorFlow, and a basic idea of a training process based on TensorFlow is that firstly, a TensorFlow session is defined; then loading historical operating data; finally, a run function of the session is invoked, minimizing the loss function. Historical operating data is required to be input into the model, and then the historical operating data is continuously subjected to cyclic training. The specific implementation process is as follows:
in one embodiment, the step S24 includes:
step S241: extracting fault characteristic data in the preprocessed historical operating data, and taking the fault characteristic data and the preprocessed historical operating data as a first initial training sample;
step S242: and sequentially filtering, complementing and standardizing the data in the first initial training sample to obtain a first target training sample.
In steps S241-S242, for example, the current, the voltage, and the coolant level constitute a historical operating data feature vector, the coolant temperature serves as a historical fault feature vector, and a first initial training sample is composed of the historical operating data feature vector and the historical fault feature vector.
In one embodiment, the z-score algorithm is used to normalize the data, and the specific formula is as follows:
where x is the actual observed value, μ is the mean of all sample data, and σ is the standard deviation of all sample data.
In one embodiment, the step S25 includes:
step S251: obtaining an LSTM network and a first target training sample;
step S252: and inputting the first target training sample into an LSTM network, and training the LSTM network by using a gradient descent method to determine a fault index prediction model.
In steps S251-S252, when the LSTM network is used to train the first target training sample, network parameters and functions need to be set, the LSTM network in the present application takes a tanh function as an activation algorithm of a hidden layer, takes cross entropy as a loss function of the LSTM network, trains the model by using a gradient descent method, and continuously updates the weight and the deviation of the model, so as to minimize the loss function. It should be noted that the operation characteristic data corresponding to the fault characteristic data in the first target training sample is input data of the LSTM network, and the fault characteristic data is output data of the LSTM network.
As shown in fig. 3, the formula of the tanh function is specifically as follows:
where tanh is a hyperbolic tangent function, the output range of tanh is between (-1,1), and the entire function is centered at 0.
In one embodiment, the gradient descent method in the present application employs an Adam algorithm, which is a first-order optimization algorithm that can replace the traditional Stochastic Gradient Descent (SGD) process, and which can iteratively update neural network weights based on training data. The Adam algorithm controls the decay rate of exponential moving means of the gradient, with the hyper-parameters beta1 and beta2 controlling the decay rate of these moving means. The initial value of the moving means and the beta1, beta2 values are close to 1 (recommended value), and therefore the deviation of the moment estimate is close to 0. The offset is improved by first calculating an estimate of the belt offset and then calculating an offset corrected estimate.
In one embodiment, the step S252 is followed by:
step S253: and recovering the fault index prediction model by using TensorFlow.
In one embodiment, the step S253 is followed by:
step S254: acquiring the fault characteristic data and a fault code corresponding to the fault characteristic data, and taking the fault characteristic data and the fault code corresponding to the fault characteristic data as a second initial training sample;
step S255: sequentially filtering, complementing and encoding the data in the second initial training sample to obtain a second target training sample;
step S256: acquiring a CNN network;
step S257: and training the CNN network by adopting the second target training sample to determine a fault classification model.
In steps S254-S257, the fault characteristic parameters and the fault codes are in a one-to-one correspondence. For example, historical operation data at a certain time point is that "voltage value (120V), current value (50MA) and coolant temperature (70 degrees centigrade)" constitute historical fault characteristics, and fault codes of historical fault data at the same time point are a, that is, "voltage value (120V), current value (50MA) and coolant temperature (70 degrees centigrade)" and fault codes a are taken as second initial training samples
In one embodiment, the present application trains CNN using the second initial training sample, and a Convolutional Neural Network (CNN) is a kind of feed forward Neural network (fed Neural network) containing convolution calculation and having a deep structure, and is one of the representative algorithms of deep learning (deep learning). The convolutional neural network is constructed by imitating a visual perception (visual perception) mechanism of an organism, supervised learning and unsupervised learning can be performed, and the sharing of convolutional kernel parameters in an implicit layer and the sparsity of connection among layers enable the convolutional neural network to be capable of carrying out grid-like topologic (grid-like) features with small calculation amount.
In an embodiment, when the second target training sample is used for training the CNN network, the operation characteristic data in the second target training sample is used as input data of the CNN network, and the fault code is used as output data of the CNN network. It should be noted that the CNN network uses sigmoid as an activation function, cross entropy as a loss function, and softmax algorithm as a probability result of network output. As shown in fig. 4, the sigmoid function is formulated as follows:
wherein the sigmoid function is continuous, smooth and strictly monotonous, is centrosymmetric about (0,0.5), the range of the value range is limited between (0,1), e is a natural constant, and x is the input of a neuron.
In one embodiment, the step S257 is followed by:
step S258: and recovering the fault classification model by using TensorFlow.
In one embodiment, the step S258 is followed by:
step S259: inputting the real-time fault characteristic index of the fuel cell system into the fault classification model, and outputting a one-hot coded probability value;
step S260: and acquiring an index value of the maximum probability value in the probability values, and acquiring the real-time fault code of the fuel cell system through inverse transformation of the one-hot code.
In steps S259-S260, when the sum of the output probability values of the fault classification model is 1, the index value of 0.72 is 0, the index value of 0.08 is 1, the index value of 0.02 is 2, and the index value of 0.18 is 3, the index with the highest probability value is 0, and the sum of the output probability values of the fault classification model is [0.72, 0.08, 0.02, 0.18 ]; one-hot codes of four fault codes of A, B, C and D are respectively [1,0,0,0], [0,1,0,0], [0,0,1,0], [0,0,0,1] and 4 fault codes are respectively 1,0,0 and 0 at the index position of 0, and the maximum value is A; therefore, the probability of [0.72, 0.08, 0.02, 0.18] is output to the corresponding fault code as a, i.e. the input feature vector index corresponds to the fault code a.
In one embodiment, the step S260 is followed by:
step S261: and carrying out persistence processing on the real-time fault code and reporting an alarm.
It should be understood that, although the steps in the flowchart of fig. 2 are shown in order as indicated by the arrows, the steps are not necessarily performed in order as indicated by the arrows. The steps are not performed in the exact order shown and described, and may be performed in other orders, unless explicitly stated otherwise. Moreover, at least a portion of the steps in fig. 2 may include multiple sub-steps or multiple stages that are not necessarily performed at the same time, but may be performed at different times, and the order of performance of the sub-steps or stages is not necessarily sequential, but may be performed in turn or alternately with other steps or at least a portion of the sub-steps or stages of other steps.
In one embodiment, as shown in fig. 5, there is provided a fuel cell system failure prediction apparatus including: a first obtaining module 10, a first processing module 20, a first determining module 30 and a second determining module 40, wherein:
the first acquisition module 10 is used for acquiring real-time operation index data of the fuel cell system;
the first processing module 20 is configured to preprocess the real-time operation index data to obtain preprocessed real-time operation index data;
a first determining module 30, configured to input the preprocessed real-time operation index data into a fault index prediction model, so as to obtain a real-time fault feature index of the fuel cell system;
and the second determining module 40 is configured to input the real-time fault feature indicator of the fuel cell system into a fault classification model, so as to obtain a real-time fault code of the fuel cell system.
In one embodiment, the first processing module 20 includes:
the first preprocessing module 201 is configured to sequentially filter, complement, and standardize the real-time operation index data to obtain preprocessed real-time operation index data.
In one embodiment, the preprocessing module 201 then comprises:
a second obtaining module 202, configured to obtain historical operating data of the fuel cell system within a preset time;
the second preprocessing module 203 is configured to preprocess the historical operating data to obtain preprocessed historical operating data;
a third determining module 204, configured to determine a first target training sample according to the preprocessed historical operating data;
a fourth determining module 205, configured to train the LSTM network by using the first target training sample, and determine a fault indicator prediction model.
In one embodiment, the third determining module 204 includes:
the data extraction module 2041 is configured to extract fault feature data in the preprocessed historical operating data, and use the fault feature data and the preprocessed historical operating data as a first initial training sample;
the third preprocessing module 2042 is configured to sequentially perform filtering, completion and standardization processing on the data in the first initial training sample, so as to obtain a first target training sample.
In one embodiment, the fourth determining module 205 includes:
a third obtaining module 2051, configured to obtain an LSTM network and a first target training sample;
the first training module 2052 is configured to input the first target training sample into an LSTM network, and train the LSTM network by using a gradient descent method to determine a fault indicator prediction model.
In one embodiment, the first training module 2052 is followed by:
a first recovery module 2053 configured to recover the failure indicator prediction model using tensrflow.
In one embodiment, the first recovery module 2053 comprises, after:
a fourth obtaining module 2054, configured to obtain the fault feature data and the fault code corresponding to the fault feature data, and use the fault feature data and the fault code corresponding to the fault feature data as a second initial training sample;
a third preprocessing module 2055, configured to sequentially filter, complement, and encode data in the second initial training sample to obtain a second target training sample;
a fifth obtaining module 2056, configured to obtain a CNN network;
and a second training module 2057, configured to train the CNN network by using the second target training sample, and determine a fault classification model.
In one embodiment, the second training module 2057 comprises, after:
a second recovery module 2058 for recovering the fault classification model using tensrflow.
In one embodiment, the second recovery module 2058 comprises, after:
a probability value determining module 2059, configured to input the real-time fault feature indicator of the fuel cell system into the fault classification model, and output a probability value of one-hot encoding;
and an inverse transformation module 2060, configured to obtain an index value of a maximum probability value in the probability values, and obtain a real-time fault code of the fuel cell system through inverse transformation of one-hot coding.
In one embodiment, the inverse transform module 2060 is followed by:
a second processing module 2061, configured to perform persistence processing on the real-time fault code and report an alarm.
For specific definition of the fuel cell system failure prediction device, reference may be made to the above definition of a fuel cell system failure prediction method, which is not described herein again. Each module in the above-described fuel cell system failure prediction apparatus may be implemented in whole or in part by software, hardware, and a combination thereof. The modules can be embedded in a hardware form or independent from a processor in the computer device, and can also be stored in a memory in the computer device in a software form, so that the processor can call and execute operations corresponding to the modules.
In one embodiment, a computer device is provided, which may be a server, and its internal structure diagram may be as shown in fig. 6. The computer device includes a processor, a memory, a network interface, and a database connected by a system bus. Wherein the processor of the computer device is configured to provide computing and control capabilities. The memory of the computer device comprises a nonvolatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program, and a database. The internal memory provides an environment for the operation of an operating system and computer programs in the non-volatile storage medium. The database of the computer device is used to store fuel cell system data. The network interface of the computer device is used for communicating with an external terminal through a network connection. The computer program is executed by a processor to implement a fuel cell system failure prediction method.
Those skilled in the art will appreciate that the architecture shown in fig. 6 is merely a block diagram of some of the structures associated with the disclosed aspects and is not intended to limit the computing devices to which the disclosed aspects apply, as particular computing devices may include more or less components than those shown, or may combine certain components, or have a different arrangement of components.
In one embodiment, a computer device is provided, comprising a memory and a processor, the memory having a computer program stored therein, the processor implementing the following steps when executing the computer program:
acquiring real-time operation index data of a fuel cell system;
preprocessing the real-time operation index data to obtain preprocessed real-time operation index data;
inputting the preprocessed real-time operation index data into a fault index prediction model to obtain a real-time fault characteristic index of the fuel cell system;
and inputting the real-time fault characteristic indexes of the fuel cell system into a fault classification model to obtain real-time fault codes of the fuel cell system.
In one embodiment, a computer-readable storage medium is provided, having a computer program stored thereon, which when executed by a processor, performs the steps of:
acquiring real-time operation index data of a fuel cell system;
preprocessing the real-time operation index data to obtain preprocessed real-time operation index data;
inputting the preprocessed real-time operation index data into a fault index prediction model to obtain a real-time fault characteristic index of the fuel cell system;
and inputting the real-time fault characteristic indexes of the fuel cell system into a fault classification model to obtain real-time fault codes of the fuel cell system.
It will be understood by those skilled in the art that all or part of the processes of the methods of the embodiments described above can be implemented by hardware instructions of a computer program, which can be stored in a non-volatile computer-readable storage medium, and when executed, can include the processes of the embodiments of the methods described above. Any reference to memory, storage, database, or other medium used in the embodiments provided herein may include non-volatile and/or volatile memory, among others. Non-volatile memory can include read-only memory (ROM), Programmable ROM (PROM), Electrically Programmable ROM (EPROM), Electrically Erasable Programmable ROM (EEPROM), or flash memory. Volatile memory can include Random Access Memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in a variety of forms such as Static RAM (SRAM), Dynamic RAM (DRAM), Synchronous DRAM (SDRAM), Double Data Rate SDRAM (DDRSDRAM), Enhanced SDRAM (ESDRAM), Synchronous Link DRAM (SLDRAM), Rambus Direct RAM (RDRAM), direct bus dynamic RAM (DRDRAM), and memory bus dynamic RAM (RDRAM).
The technical features of the above embodiments can be arbitrarily combined, and for the sake of brevity, all possible combinations of the technical features in the above embodiments are not described, but should be considered as the scope of the present specification as long as there is no contradiction between the combinations of the technical features.
The above-mentioned embodiments only express several embodiments of the present application, and the description thereof is more specific and detailed, but not construed as limiting the scope of the invention. It should be noted that, for a person skilled in the art, several variations and modifications can be made without departing from the concept of the present application, which falls within the scope of protection of the present application. Therefore, the protection scope of the present patent shall be subject to the appended claims.
Claims (13)
1. A fuel cell system failure prediction method, characterized in that the method comprises:
acquiring real-time operation index data of a fuel cell system;
preprocessing the real-time operation index data to obtain preprocessed real-time operation index data;
inputting the preprocessed real-time operation index data into a fault index prediction model to obtain a real-time fault characteristic index of the fuel cell system;
and inputting the real-time fault characteristic indexes of the fuel cell system into a fault classification model to obtain real-time fault codes of the fuel cell system.
2. The method of claim 1, wherein the preprocessing the real-time operation index data to obtain preprocessed real-time operation index data comprises:
and sequentially filtering, complementing and standardizing the real-time operation index data to obtain the preprocessed real-time operation index data.
3. The method according to claim 2, wherein the sequentially filtering, complementing and standardizing the real-time operation index data to obtain the preprocessed real-time operation index data comprises:
acquiring historical operating data of a fuel cell system within preset time;
preprocessing the historical operating data to obtain preprocessed historical operating data;
determining a first target training sample according to the preprocessed historical operating data;
and training the LSTM network by adopting the first target training sample to determine a fault index prediction model.
4. The method of claim 3, wherein determining a first target training sample based on the pre-processed historical operating data comprises:
extracting fault characteristic data in the preprocessed historical operating data, and taking the fault characteristic data and the preprocessed historical operating data as a first initial training sample;
and sequentially filtering, complementing and standardizing the data in the first initial training sample to obtain a first target training sample.
5. The method of claim 4, wherein training the LSTM network using the first target training samples to determine a fault indicator prediction model comprises:
obtaining an LSTM network and a first target training sample;
and inputting the first target training sample into an LSTM network, and training the LSTM network by using a gradient descent method to determine a fault index prediction model.
6. The method of claim 5, wherein the inputting the first target training sample into an LSTM network and training the LSTM network using a gradient descent method to determine the fault indicator prediction model comprises:
and recovering the fault index prediction model by using TensorFlow.
7. The method according to claim 6, wherein the recovering the fault metric predictive model using the TensorFlow is followed by:
acquiring the fault characteristic data and a fault code corresponding to the fault characteristic data, and taking the fault characteristic data and the fault code corresponding to the fault characteristic data as a second initial training sample;
sequentially filtering, complementing and encoding the data in the second initial training sample to obtain a second target training sample;
acquiring a CNN network;
and training the CNN network by adopting the second target training sample to determine a fault classification model.
8. The method of claim 7, wherein the training the CNN network using the second target training sample, and determining the fault classification model comprises:
and recovering the fault classification model by using TensorFlow.
9. The method according to claim 8, wherein the recovering the fault classification model using the TensorFlow comprises, after:
inputting the real-time fault characteristic index of the fuel cell system into the fault classification model, and outputting a one-hot coded probability value;
and acquiring an index value of the maximum probability value in the probability values, and acquiring the real-time fault code of the fuel cell system through inverse transformation of the one-hot code.
10. The method of claim 9, wherein obtaining the index value of the maximum probability value of the probability values after obtaining the real-time fault code of the fuel cell system through inverse transformation of one-hot coding comprises:
and carrying out persistence processing on the real-time fault code and reporting an alarm.
11. A fuel cell system failure prediction apparatus, characterized in that the apparatus comprises:
the first acquisition module is used for acquiring real-time operation index data of the fuel cell system;
the first processing module is used for preprocessing the real-time operation index data to obtain preprocessed real-time operation index data;
the first determination module is used for inputting the preprocessed real-time operation index data into a fault index prediction model to obtain a real-time fault characteristic index of the fuel cell system;
and the second determination module is used for inputting the real-time fault characteristic indexes of the fuel cell system into a fault classification model to obtain real-time fault codes of the fuel cell system.
12. A computer device comprising a memory and a processor, the memory storing a computer program, characterized in that the processor realizes the steps of the method of any one of claims 1 to 10 when executing the computer program.
13. A computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, carries out the steps of the method of any one of claims 1 to 10.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911196734.0A CN112882909A (en) | 2019-11-29 | 2019-11-29 | Fuel cell system fault prediction method and device |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201911196734.0A CN112882909A (en) | 2019-11-29 | 2019-11-29 | Fuel cell system fault prediction method and device |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112882909A true CN112882909A (en) | 2021-06-01 |
Family
ID=76038352
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201911196734.0A Pending CN112882909A (en) | 2019-11-29 | 2019-11-29 | Fuel cell system fault prediction method and device |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112882909A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113282433A (en) * | 2021-06-10 | 2021-08-20 | 中国电信股份有限公司 | Cluster anomaly detection method and device and related equipment |
| CN113361823A (en) * | 2021-07-16 | 2021-09-07 | 同济大学 | Fuel cell fault prediction method and system based on prediction data |
| CN115805810A (en) * | 2022-10-31 | 2023-03-17 | 宁德时代新能源科技股份有限公司 | Battery failure prediction method, apparatus, device, storage medium, and program product |
| CN116519021A (en) * | 2023-06-29 | 2023-08-01 | 西北工业大学 | Inertial navigation system fault diagnosis method, system and equipment |
| CN117540199A (en) * | 2024-01-05 | 2024-02-09 | 中国汽车技术研究中心有限公司 | Fault prediction method, device and storage medium for fuel cell vehicle |
-
2019
- 2019-11-29 CN CN201911196734.0A patent/CN112882909A/en active Pending
Cited By (7)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113282433A (en) * | 2021-06-10 | 2021-08-20 | 中国电信股份有限公司 | Cluster anomaly detection method and device and related equipment |
| CN113361823A (en) * | 2021-07-16 | 2021-09-07 | 同济大学 | Fuel cell fault prediction method and system based on prediction data |
| CN115805810A (en) * | 2022-10-31 | 2023-03-17 | 宁德时代新能源科技股份有限公司 | Battery failure prediction method, apparatus, device, storage medium, and program product |
| CN116519021A (en) * | 2023-06-29 | 2023-08-01 | 西北工业大学 | Inertial navigation system fault diagnosis method, system and equipment |
| CN116519021B (en) * | 2023-06-29 | 2023-09-15 | 西北工业大学 | Inertial navigation system fault diagnosis method, system and equipment |
| CN117540199A (en) * | 2024-01-05 | 2024-02-09 | 中国汽车技术研究中心有限公司 | Fault prediction method, device and storage medium for fuel cell vehicle |
| CN117540199B (en) * | 2024-01-05 | 2024-05-07 | 中国汽车技术研究中心有限公司 | Fault prediction method, device and storage medium for fuel cell vehicle |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN112882909A (en) | Fuel cell system fault prediction method and device | |
| CN112131673B (en) | Engine surge fault prediction system and method based on fusion neural network model | |
| Laredo et al. | A neural network-evolutionary computational framework for remaining useful life estimation of mechanical systems | |
| CN112907781B (en) | System fault diagnosis method and device, computer equipment and storage medium | |
| CN113762486B (en) | Method and device for constructing fault diagnosis model of converter valve and computer equipment | |
| Liao et al. | Combining deep learning and survival analysis for asset health management | |
| CN112132394B (en) | Power plant circulating water pump predictive state evaluation method and system | |
| CN113760670A (en) | Cable joint abnormity early warning method and device, electronic equipment and storage medium | |
| Kara | Multi-scale deep neural network approach with attention mechanism for remaining useful life estimation | |
| Kefalas et al. | Automated machine learning for remaining useful life estimation of aircraft engines | |
| Shcherbakov et al. | A hybrid deep learning framework for intelligent predictive maintenance of cyber-physical systems | |
| CN112766618A (en) | Anomaly prediction method and device | |
| CN113807951A (en) | Transaction data trend prediction method and system based on deep learning | |
| CN116383722A (en) | Fracturing measure process monitoring method based on gate control circulation unit neural network | |
| Lin et al. | Remaining useful life prediction in prognostics using multi-scale sequence and Long Short-Term Memory network⋆ | |
| Xu et al. | Global attention mechanism based deep learning for remaining useful life prediction of aero-engine | |
| Chen et al. | Multivariate time series anomaly detection via dynamic graph forecasting | |
| de Miranda et al. | Recurrent neural network based on statistical recurrent unit for remaining useful life estimation | |
| Li et al. | Knowledge enhanced ensemble method for remaining useful life prediction under variable working conditions | |
| Vishnu et al. | Data-driven prognostics with predictive uncertainty estimation using ensemble of deep ordinal regression models | |
| CN115616408A (en) | Battery thermal management data processing method and system | |
| CN115017819A (en) | Engine remaining service life prediction method and device based on hybrid model | |
| CN114429246A (en) | Power grid potential safety determination method and device based on heterogeneous neural network | |
| Fu et al. | Lithium-ion battery SOH prediction based on VMD-PE and improved DBO optimized temporal convolutional network model | |
| Singh et al. | Deep machine learning approaches for battery health monitoring |
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 |