CN115308608A - All-vanadium redox flow battery voltage prediction method, device and medium - Google Patents
All-vanadium redox flow battery voltage prediction method, device and medium Download PDFInfo
- Publication number
- CN115308608A CN115308608A CN202210921665.0A CN202210921665A CN115308608A CN 115308608 A CN115308608 A CN 115308608A CN 202210921665 A CN202210921665 A CN 202210921665A CN 115308608 A CN115308608 A CN 115308608A
- Authority
- CN
- China
- Prior art keywords
- redox flow
- vanadium redox
- flow battery
- data
- soc
- 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
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/367—Software therefor, e.g. for battery testing using modelling or look-up tables
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/3644—Constructional arrangements
- G01R31/3648—Constructional arrangements comprising digital calculation means, e.g. for performing an algorithm
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/378—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC] specially adapted for the type of battery or accumulator
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/385—Arrangements for measuring battery or accumulator variables
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01R—MEASURING ELECTRIC VARIABLES; MEASURING MAGNETIC VARIABLES
- G01R31/00—Arrangements for testing electric properties; Arrangements for locating electric faults; Arrangements for electrical testing characterised by what is being tested not provided for elsewhere
- G01R31/36—Arrangements for testing, measuring or monitoring the electrical condition of accumulators or electric batteries, e.g. capacity or state of charge [SoC]
- G01R31/389—Measuring internal impedance, internal conductance or related variables
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02E—REDUCTION OF GREENHOUSE GAS [GHG] EMISSIONS, RELATED TO ENERGY GENERATION, TRANSMISSION OR DISTRIBUTION
- Y02E60/00—Enabling technologies; Technologies with a potential or indirect contribution to GHG emissions mitigation
- Y02E60/30—Hydrogen technology
- Y02E60/50—Fuel cells
Landscapes
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Fuel Cell (AREA)
Abstract
The invention provides a method, a device and a medium for predicting the voltage of an all-vanadium redox flow battery, which are used for fitting experimental data of the all-vanadium redox flow battery based on a gated cyclic neural network model and predicting the voltage of the all-vanadium redox flow battery, and the method comprises the following steps: s10, collecting all-vanadium redox flow battery experimental data; s20, calculating the SOC of the all-vanadium redox flow battery under different working conditions according to experimental data to form a training data set; s30, constructing a gated cyclic neural network model, and training the gated cyclic neural network model through a training data set; and collecting the current, the capacity and the flow of the battery to be tested, calculating the current SOC data, and inputting the current SOC data into the trained gated cyclic neural network model to obtain a predicted voltage value. According to the method, the flow is considered to be an important factor influencing the operation efficiency of the all-vanadium redox flow battery, the flow is added as the input of the model, the prediction precision of the model under different flows is further improved, the model is simple, and the prediction error is small.
Description
Technical Field
The invention relates to the technical field of all-vanadium redox flow batteries, in particular to a method, a device and a medium for predicting the voltage of an all-vanadium redox flow battery.
Background
The Vanadium Redox Flow Battery (VRB) is a new electrochemical energy storage technology, and has the advantages of flexible system design, large storage capacity, safety, environmental protection, long service life and the like.
However, the existing voltage prediction method of the all-vanadium redox flow battery is mainly realized by depending on an electrochemical model and an equivalent circuit model, wherein the electrochemical model is suitable for demand scenes such as battery principle analysis and battery research and development, and the equivalent circuit model is usually suitable for engineering practice fields such as operation control and dynamic response.
However, the above prediction method needs to understand the internal principle of the model, and simplifies the output voltage of the battery into a simple transformation superposition of a linear function and a negative exponential function, so that on one hand, the model is complex, and the fitting capability to the nonlinear behavior of the battery is limited. On the other hand, the influence of the flow speed on the battery voltage is not considered, and the prediction effect is poor.
Disclosure of Invention
Aiming at the problems in the background art, the invention provides a method, a device and a medium for predicting the voltage of an all-vanadium redox flow battery, which can realize high-precision prediction of the output characteristic of the all-vanadium redox flow battery, and have the advantages of simple model and good nonlinear fitting effect.
In order to achieve the purpose, the invention provides a voltage prediction method for an all-vanadium redox flow battery, which is characterized by comprising the following steps:
s1, collecting experimental data of the all-vanadium redox flow battery, wherein the experimental data comprises historical voltage, current, capacity and flow of the all-vanadium redox flow battery;
s2, calculating the SOC of the all-vanadium redox flow battery under different working conditions according to the experimental data, and preprocessing the experimental data and the corresponding SOC data to form a training data set;
s3, constructing a gated cyclic neural network model, and training the gated cyclic neural network model through a training data set; and collecting the current, the capacity and the flow of the battery to be tested, calculating the current SOC data, and inputting the current SOC data into the trained gated cyclic neural network model to obtain a predicted voltage value.
Preferably, in step S1, experimental data of the all-vanadium redox flow battery is collected by a monitoring module, the monitoring module controls a charge and discharge experiment of the all-vanadium redox flow battery, and simultaneously collects external characteristics of the all-vanadium redox flow battery and stores the external characteristics as experimental data.
Preferably, the specific steps of step S3 include:
s202, acquiring an SOC-OCV relation curve of the all-vanadium redox flow battery system, and determining a corresponding relation between the state of charge and the open-circuit voltage of the all-vanadium redox flow battery;
s204, acquiring open-circuit voltage of the all-vanadium redox flow battery in no-load based on the SOC-OCV relation curve of the all-vanadium redox flow battery system, and reading corresponding initial SOC data;
s206, calculating current SOC data based on the capacity and the initial SOC data in the experimental data;
s208, acquiring the experimental data and the SOC data;
s210, normalization processing is carried out on the experimental data and the SOC data, and a training data set is generated.
Preferably, the method for obtaining the SOC-OCV relationship curve of the all-vanadium redox flow battery system in step S202 is based on a low-current constant-current charge and discharge experiment, and after the open-circuit voltage OCV and the battery SOC value at each moment are recorded in the experiment process, a corresponding SOC-OCV relationship curve is drawn.
Preferably, the calculation formula for calculating the current SOC data in step S206 is:
wherein, SOC charging 、SOC discharging Respectively the current SOC data and SOC during charging and discharging 0 As initial SOC data, SOC total_charging Capicity being the total SOC data accumulate For volumetric data, capicity nominal_charging Is rated capacity.
Preferably, the specific steps of step S3 include:
step S302: constructing a gated cyclic neural network model, wherein the gated cyclic neural network model comprises an input layer, a hidden layer and an output layer, the input layer is connected with the output layer through the hidden layer, the training data set enters the gated cyclic neural network model through the input layer for fitting calculation, and the input layer reads the training data set and sends the training data set to the hidden layer;
step S304: the hidden layer fits the training data set, optimizes iteration parameter values and generates an output voltage limit interval;
step S306: and the output layer performs data calculation based on the received training data set, the output voltage limit interval and the iteration parameter value to predict the voltage value of the all-vanadium redox flow battery.
Preferably, the hidden layer includes a gated cycle unit layer, an activation function layer, and a full connection layer, the gated cycle unit layer is connected to the output layer via the activation unit layer and the full connection layer, respectively, the hidden layer fits the training data set, optimizes iteration parameter values, and generates an output voltage limit interval, and the specific steps include:
the gating cycle unit layer extracts the time sequence characteristics of the training data set;
the activation unit layer reads the training data set to generate the output voltage limit interval;
and the full connection layer performs iterative computation based on the time sequence characteristics, the output voltage limit interval and the SOC data to generate an optimized iterative parameter value.
Preferably, the fully-connected layer includes an error layer and a correction layer, wherein:
the error layer calculates a fitting error based on a loss function, the acquired time sequence characteristics, the output voltage limit interval and SOC data;
and if the fitting error is greater than or equal to a preset error threshold value, the correction layer generates the optimized iteration parameter.
The invention also provides a device for predicting the voltage of the all-vanadium redox flow battery, which is characterized by comprising the following components:
a data acquisition module: the system is used for acquiring experimental data of the all-vanadium redox flow battery acquired by a monitoring system, and the all-vanadium redox flow battery is connected with an upper computer through the monitoring module;
a data processing module: the system comprises a data acquisition module, a data storage module and a data processing module, wherein the data acquisition module is connected with the data acquisition module and is used for calculating the SOC of the all-vanadium redox flow battery under different working conditions according to the experimental data and preprocessing the experimental data and the corresponding SOC data to form a training data set;
and the neural network module is connected with the data processing module and used for performing parameter fitting calculation by adopting a gated cyclic neural network model based on the training data set so as to predict the voltage value of the all-vanadium redox flow battery.
The invention further proposes a computer-readable storage medium, in which a computer program is stored, which is characterized in that the computer program realizes the steps of the above-mentioned method when being executed by a processor.
The invention has the beneficial effects that:
1. according to the method, the gated cyclic neural network model is adopted to perform fitting calculation based on the training data set so as to predict the voltage value of the all-vanadium redox flow battery, the extraction and analysis of the battery time sequence characteristics are realized, and the nonlinear prediction capability of the model is improved.
2. According to the method, the flow is considered to be an important factor influencing the operation efficiency of the all-vanadium redox flow battery, the flow is added as the input of the model, the prediction precision of the model under different flows is further improved, the model is simple, and the prediction error is small.
3. The all-vanadium redox flow battery voltage prediction method provided by the invention can be effectively applied to engineering practice, and the prediction precision of the model is improved.
Drawings
FIG. 1 is a schematic flow chart of a voltage prediction method of an all-vanadium redox flow battery according to the present invention;
fig. 2 is a schematic flowchart of a voltage prediction method for an all-vanadium redox flow battery according to an embodiment of the present invention;
FIG. 3 is a schematic structural diagram of a gated recurrent neural network model according to an embodiment of the present invention;
FIG. 4 is a schematic diagram illustrating a hidden layer in a gated recurrent neural network model according to an embodiment of the present invention;
fig. 5 is a schematic flowchart of a voltage prediction method for an all-vanadium redox flow battery according to another embodiment of the present invention;
fig. 6 is a schematic diagram illustrating the effect of fitting the experimental data of the all-vanadium redox flow battery based on a gated cycle network model according to another embodiment of the present invention;
fig. 7 is a schematic structural diagram of an all-vanadium redox flow battery voltage prediction apparatus according to an embodiment of the present invention;
in the figure, the all-vanadium redox flow battery voltage prediction device 100, the data acquisition module 10, the data processing module 20 and the neural network module 30 are shown.
Detailed Description
To facilitate an understanding of the present application, the present application will now be described more fully with reference to the accompanying drawings. Preferred embodiments of the present application are illustrated in the accompanying drawings. This application may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete.
Unless defined otherwise, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art to which this application belongs. The terminology used herein in the description of the present application is for the purpose of describing particular embodiments only and is not intended to be limiting of the application. As used herein, the term "and/or" includes any and all combinations of one or more of the associated listed items.
Where the terms "comprising," "having," and "including" are used herein, another component can be added unless an explicit limitation is used, such as "only," "consisting of … …," and the like. Unless mentioned to the contrary, terms in the singular may include the plural and are not to be construed as being one in number.
It will be understood that, although the terms first, second, etc. may be used herein to describe various elements, these elements should not be limited by these terms. These terms are only used to distinguish one element from another. For example, a first element could be termed a second element, and, similarly, a second element could be termed a first element, without departing from the scope of the present application.
In this application, unless otherwise expressly stated or limited, the terms "connected" and "coupled" are to be construed broadly and encompass, for example, direct and indirect coupling via an intermediary, and communication between two elements or an interaction between two elements. The specific meaning of the above terms in the present application can be understood by those of ordinary skill in the art as the case may be.
The all-vanadium redox flow battery is a liquid redox battery which takes vanadium as an active substance and flows circularly. From the generation of all-vanadium redox flow batteries to date, expert scholars at home and abroad have researched a plurality of widely known battery models. Starting from the constructed principle, the all-vanadium redox flow battery model is mainly divided into an electrochemical model and an equivalent circuit model.
The electrochemical model of the vanadium redox battery can comprehensively reflect the characteristics of the vanadium redox battery, but when parameter estimation is carried out, the involved mathematical equation operation is complex, and a diffusion phenomenon can occur during experiment, so that the electrochemical model is difficult to be applied to actual engineering. The equivalent circuit model of the vanadium redox battery is a model provided by combining the volt-ampere characteristic and the internal loss of the vanadium redox battery, has the characteristic of nonlinearity, ignores the mutual restriction relationship among the chemical reaction in the battery, the loss of each part including a pump and a galvanic pile, the ion movement and the concentration, and cannot realize the high-precision prediction of the output characteristic of the battery.
Aiming at the problems, the application provides a voltage prediction method for the all-vanadium redox flow battery, which can be effectively applied to engineering practice and can improve the prediction precision of a model. To illustrate this technical solution, the following is a description with specific examples.
The invention provides an all-vanadium redox flow battery voltage prediction method, which is used for fitting experimental data of an all-vanadium redox flow battery based on a gated cyclic neural network model and predicting the voltage of the all-vanadium redox flow battery, and as shown in figure 1, the method comprises the following steps:
s1, collecting experimental data of the all-vanadium redox flow battery, wherein the experimental data comprises historical voltage, current, capacity and flow of the all-vanadium redox flow battery;
s2, calculating the SOC of the all-vanadium redox flow battery under different working conditions according to the experimental data, and preprocessing the experimental data and the corresponding SOC data to form a training data set;
s3, constructing a gated cyclic neural network model, and training the gated cyclic neural network model through a training data set; and collecting the current, the capacity and the flow of the battery to be tested, calculating the current SOC data, and inputting the current SOC data into the trained gated cyclic neural network model to obtain a predicted voltage value.
In one embodiment, step S10: the method comprises the steps of obtaining experimental data monitored by a monitoring module, connecting the all-vanadium redox flow battery with an upper computer through the monitoring module, collecting the experimental data of the all-vanadium redox flow battery through the monitoring module, controlling the charge and discharge experiment of the all-vanadium redox flow battery through the monitoring module, and simultaneously collecting the external characteristics of the all-vanadium redox flow battery and storing the external characteristics as the experimental data. The monitoring module comprises a battery performance tester (BTS-5V 6A), the tester comprises a lower computer and a middle computer, the lower computer is connected with the battery to be tested and used for collecting the external characteristics of the battery, storing the external characteristics as experimental data, receiving a control instruction of the middle computer to control the charging and discharging experiment of the battery to be tested, and the middle computer receives the control instruction of the upper computer and transmits the experimental data.
Specifically, the experimental data comprises historical voltage, current, capacity and flow of the all-vanadium redox flow battery. Due to the fact that the operation mechanism of the all-vanadium redox flow battery is complex, internal parameters of the battery can be changed under different battery operation conditions, such as changes of working conditions of charging and discharging current, flow and the like, so that external characteristics of the battery are influenced, prediction accuracy of a model is influenced, the flow is used as a specific attribute of the all-vanadium redox flow battery, and the concentration of vanadium ions and the concentration of protons in the battery are influenced by the flow of electrolyte, so that voltage of a galvanic pile is influenced. According to the prediction method, experimental data are collected in multiple dimensions, fitting calculation is carried out, and prediction accuracy of the model is effectively improved.
Step S20: and preprocessing the experimental data to generate a training data set. Because the experimental data are huge, a training data set is generated for model training after the data required by the prediction model are subjected to screening pretreatment.
The State of Charge (SOC) reflects a ratio of a schedulable energy storage capacity of an energy storage system to a maximum available energy storage capacity at any time, and is an important parameter for representing the performance of the all-vanadium redox flow battery, and specifically, the embodiment provides a SOC data calculation method, as shown in fig. 2, the specific steps include:
s202, an SOC-OCV relation curve of the all-vanadium redox flow battery system is obtained, and a corresponding relation between the state of charge and the open-circuit voltage of the all-vanadium redox flow battery is determined.
In the initial charging and discharging stage of the all-vanadium redox flow battery, the battery capacity cannot be completely released or fully charged, so that the initial SOC of the battery and the OCV of the battery have a linear relationship, so that SOC calibration can be realized by obtaining an SOC-OCV relationship curve, and based on a small current experiment, because ohmic resistance partial pressure exists in the experiment process, a monitoring device can only measure the terminal voltage of the battery, and the OCV of the battery cannot be directly measured.
S204, acquiring the open-circuit voltage of the all-vanadium redox flow battery in no-load based on the SOC-OCV relation curve of the all-vanadium redox flow battery system, and reading corresponding initial SOC data.
Before starting an all-vanadium redox flow battery experiment under different working conditions, the capacity of the battery needs to be calibrated, specifically, at the initial stage of the experiment, the OCV during no-load is obtained, and the calibrated SOC is obtained corresponding to the SOC-OCV relation curve under a small current experiment 0 I.e., initial SOC data.
S206 calculates current SOC data based on the capacity and the initial SOC data in the experimental data.
The calculation formula is as follows:
therein, SOC charging 、SOC discharging Respectively the current SOC data and SOC during charging and discharging 0 As initial SOC data, SOC total_charging Capicity as total SOC data accumulate For volumetric data, capicity nominal_charging Is rated capacity.
S208, acquiring the experimental data and the SOC data.
S210, normalization processing is carried out on the experimental data and the SOC data, and a training data set is generated.
The normalization method employed in this embodiment is linear normalization, and defines the input value between [0,1] by linear variation:
wherein x max And x min Maximum and minimum values, x, in the data sequence, respectively i To input a numerical value, x i_nirmalization And normalizing each element in the experimental data and the SOC data to generate a training data set for the normalized data value.
Step S30: and performing fitting calculation by using a gated cyclic neural network model based on the training data set to predict the voltage value of the all-vanadium redox flow battery.
In one embodiment, as shown in fig. 3, the gated recurrent neural network model includes an input layer, a hidden layer, and an output layer, the input layer is connected to the output layer via the hidden layer, specifically, the hidden layer structure is as shown in fig. 4, the gated recurrent neural network model can selectively memorize the key information in front of the sequence, so as to improve the data processing capability of the model, and the state update manner is as follows:
wherein h is t Is the state at the current time, h t-1 Is the state at the previous moment in time,as candidate state at the current time, x t An input of the current time, an.
By updating the door z t ∈[0,1]To control the balance between input and forgetting:
z t =σ(W z x t +U z h t-1 +b z ) (ii) a Wherein, W z 、U z 、b z Are learnable network parameters.
σ (-) is the Logistic function with an output interval of (0,1).
When z is t H when =0 t And h t-1 Is a nonlinear functional relation; when z is t H when =1 t And h t-1 Is a linear function relationship.
By resetting the gate r t ∈[0,1]To control the candidate stateWhether the state at the previous moment is relied on is specifically shown as the following formula:
when r is t When the value is not less than 0, the reaction time is not less than 0,dependent only on x t Independent of historical state; when r is t When the pressure is not greater than 1, the pressure is lower than 1,depends on x t And h t-1 。
As shown in fig. 5, the specific steps of S30 include:
s302, a gated cyclic neural network model is built, the training data set enters the gated cyclic neural network model through an input layer to perform fitting calculation, and the input layer reads the training data set and sends the training data set to the hidden layer.
Specifically, as an example, the input layer reads a training data set x 1:T =(x 1 ,x 2 ,…,x t ,…x T ) And sending to the hidden layer.
S304, fitting the training data set by the hidden layer, optimizing iterative parameter values and generating an output voltage limit interval;
specifically, the hidden layer receives x 1:T =(x 1 ,x 2 ,…,x t ,…x T ) And updating the formula through optimization: h is t =f(h t-1 ,x t ) And updating the hidden layer iteration parameter value. And to be deliveredAnd limiting the output result to generate an output voltage limit interval.
S306, the output layer carries out data calculation based on the received training data set, the output voltage limit interval and the iteration parameter value so as to predict the voltage value of the all-vanadium redox flow battery.
Specifically, input data are continuously subjected to iterative calculation through a gated cyclic neural network model, and finally the voltage value of the all-vanadium redox flow battery is output as a prediction.
In one embodiment, the hidden layer comprises a gated cyclic unit layer, an activation function layer and a fully connected layer, the gated cyclic unit layer being connected to the output layer via the activation unit layer and the fully connected layer, respectively.
S3062, extracting time sequence characteristics of the training data set based on the gating circulation unit layer; the external characteristics of the battery show regularity and continuity along with time change, and the long-term dependence problem of the recurrent neural network is improved based on the time sequence characteristics of the gated cyclic unit layer extracted training set, so that the method is more suitable for voltage prediction of the battery.
S3064 reading the training data set based on the activation unit layer to generate the output voltage limit interval;
specifically, in the battery experiment, the charge-discharge cutoff voltage is set artificially, and data points including the charge-discharge cutoff voltage are few, so that the neural network can hardly learn the cutoff condition of the voltage; however, by adding a layer of activation cells (sigmoid layer) to the neural network, the output voltage limit interval is generated, for example: the output of the neural network is limited in the range of [0,1], and the denormalized prediction result can basically meet the requirement of prediction accuracy.
S3066, the full connection layer carries out iterative calculation based on the time sequence characteristics, the output voltage limit interval and the voltage value of the all-vanadium redox flow battery to generate optimized iterative parameter values.
Specifically, when the full connection layer obtains a predicted battery voltage value, it needs to continuously generate an optimized iteration parameter value based on information such as a time sequence characteristic and a voltage limit interval, so that model prediction is more accurate.
In one embodiment, the fully-connected layer includes an error layer and a correction layer, and the method includes the following steps:
s30662 the error layer calculates a fitting error based on the loss function, the acquired time sequence characteristics, the output voltage limit interval and the voltage value of the all-vanadium redox flow battery;
specifically, the loss function employed is the Mean Square Error (MSE):
the MSE ranges from [0, + ∞ ] and the larger the error, the larger the value, the worse the model performance.
S30664, if the fitting error is greater than or equal to a preset error threshold, the correction layer generates the optimized iteration parameter.
Specifically, a model fitting error is calculated based on the loss function, if the error value is larger than a preset error value, the model optimizes iteration parameters through a correction layer, and the next iteration calculation is carried out until the error value meets the expected requirement.
In the all-vanadium redox flow battery voltage prediction method in the embodiment, the experimental data of the all-vanadium redox flow battery is fitted based on a gated cyclic neural network model to predict the battery voltage, as an example, referring to fig. 6, the model is trained by taking constant-flow variable-current experimental data under one charge-discharge cycle as a training set, and the single-cycle constant-flow constant-current experimental data is predicted, wherein a dotted line represents a predicted voltage value, a solid line represents an actual voltage value, and obviously, the predicted value and the actual value of the model are relatively close to each other in trend, and the prediction effect is relatively good; in the end stage of charging and discharging, because the charging and discharging cutoff voltage is set for people, data points containing the charging and discharging cutoff voltage are few, and therefore the neural network can hardly learn the cutoff condition of the voltage; however, by adding a sigmoid layer in the neural network, the output of the neural network is limited in the range of [0,1], and the prediction result after denormalization basically meets the requirement of prediction precision. The prediction method provided by the invention can effectively improve the nonlinear prediction capability of the all-vanadium redox flow battery model, and meanwhile, the flow is added as the input of the model in consideration of the fact that the flow is an important factor influencing the operation efficiency of the all-vanadium redox flow battery, so that the prediction accuracy of the model under different flows is further improved.
The application provides an all-vanadium redox flow battery voltage prediction device 100, which is used for fitting experimental data of an all-vanadium redox flow battery based on a gated recurrent neural network model, and the device is shown in fig. 7 and comprises the following components: a data acquisition module 10, a data processing module 20 and a neural network module 30.
The data acquisition module 10: the all-vanadium redox flow battery monitoring system is used for acquiring experimental data of the all-vanadium redox flow battery monitored and acquired by the monitoring system, and the all-vanadium redox flow battery is connected with an upper computer through the monitoring module;
a data processing module: the data acquisition module 10 is connected with the all-vanadium redox flow battery and is used for calculating the SOC of the all-vanadium redox flow battery under different working conditions according to the experimental data and preprocessing the experimental data and the corresponding SOC data to form a training data set;
and the neural network module 30 is connected with the data processing module 20 and is used for performing parameter fitting calculation by using a gated cyclic neural network model based on the training data set so as to predict the voltage value of the all-vanadium redox flow battery.
According to the voltage prediction device for the all-vanadium redox flow battery in the embodiment, the data acquisition module, the data processing module and the neural network module are arranged to be matched with each other, experimental data are fitted based on the gated cyclic neural network model, and the time sequence characteristics of the experimental data are extracted to predict the voltage of the battery. The device can effectively improve the nonlinear prediction capability of the all-vanadium redox flow battery model, and is suitable for the field of engineering practice.
In an embodiment of the present application, a computer device is further proposed, which includes a memory and a processor, the memory storing a computer program, and the processor implementing the steps of the method described in any of the embodiments of the present application when executing the computer program.
In an embodiment of the present application, a computer-readable storage medium is further proposed, on which a computer program is stored, which computer program, when being executed by a processor, is adapted to carry out the steps of the method described in any of the embodiments of the present application.
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 can include at least one of non-volatile and volatile memory. Non-volatile Memory may include Read-Only Memory (ROM), magnetic tape, floppy disk, flash Memory, optical storage, or the like. Volatile Memory can include Random Access Memory (RAM) or external cache Memory. By way of illustration and not limitation, RAM can take many forms, such as Static Random Access Memory (SRAM) or Dynamic Random Access Memory (DRAM), among others.
It should be noted that the above-mentioned embodiments are only for illustrative purposes and are not meant to limit the present invention.
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 (10)
1. A method for predicting the voltage of an all-vanadium redox flow battery is characterized by comprising the following steps: the method comprises the following steps:
s10, collecting experimental data of the all-vanadium redox flow battery, wherein the experimental data comprises historical voltage, current, capacity and flow of the all-vanadium redox flow battery;
s20, calculating the SOC of the all-vanadium redox flow battery under different working conditions according to the experimental data, and preprocessing the experimental data and the corresponding SOC data to form a training data set;
s30, constructing a gated cyclic neural network model, and training the gated cyclic neural network model through a training data set; and collecting the current, the capacity and the flow of the battery to be tested, calculating the current SOC data, and inputting the current SOC data into the trained gated cyclic neural network model to obtain a predicted voltage value.
2. The all-vanadium redox flow battery voltage prediction method according to claim 1, characterized in that: in the step S10, experimental data of the all-vanadium redox flow battery are collected through a monitoring module, the monitoring module controls a charge-discharge experiment of the all-vanadium redox flow battery, and meanwhile, external characteristics of the all-vanadium redox flow battery are collected and stored as experimental data.
3. The all-vanadium redox flow battery voltage prediction method according to claim 1, characterized in that: the specific steps of step S20 include:
s202, acquiring an SOC-OCV relation curve of the all-vanadium redox flow battery system, and determining a corresponding relation between the state of charge and the open-circuit voltage of the all-vanadium redox flow battery;
s204, acquiring open-circuit voltage of the all-vanadium redox flow battery in no-load based on the SOC-OCV relation curve of the all-vanadium redox flow battery system, and reading corresponding initial SOC data;
s206, calculating current SOC data based on the capacity and the initial SOC data in the experimental data;
s208, acquiring the experimental data and the SOC data;
s210, normalization processing is carried out on the experimental data and the SOC data, and a training data set is generated.
4. The all-vanadium redox flow battery voltage prediction method according to claim 3, characterized in that: the method for obtaining the SOC-OCV relation curve of the all-vanadium redox flow battery system in the step S202 is based on a low-current constant-current charging and discharging experiment, and after an open-circuit voltage OCV and a battery SOC value at each moment are recorded in the experiment process, a corresponding SOC-OCV relation curve is drawn.
5. The all-vanadium redox flow battery voltage prediction method according to claim 3, characterized in that: the calculation formula for calculating the current SOC data in step S206 is:
therein, SOC charging 、SOC discharging Respectively the current SOC data and SOC during charging and discharging 0 As initial SOC data, SOC total_charging Capicity being the total SOC data accumulate For volumetric data, capicity nominal_charging Is rated capacity.
6. The all-vanadium redox flow battery voltage prediction method according to claim 1, characterized in that: the specific steps of step S30 include:
step S302: constructing a gated cyclic neural network model, wherein the gated cyclic neural network model comprises an input layer, a hidden layer and an output layer, the input layer is connected with the output layer through the hidden layer, the training data set enters the gated cyclic neural network model through the input layer for fitting calculation, and the input layer reads the training data set and sends the training data set to the hidden layer;
step S304: the hidden layer fits the training data set, optimizes iteration parameter values and generates an output voltage limit interval;
step S306: and the output layer performs data calculation based on the received training data set, the output voltage limit interval and the iteration parameter value to predict the voltage value of the all-vanadium redox flow battery.
7. The all-vanadium redox flow battery voltage prediction method according to claim 6, characterized in that: the hidden layer comprises a gate control circulation unit layer, an activation function layer and a full connection layer, the gate control circulation unit layer is connected with the output layer through the activation unit layer and the full connection layer, the hidden layer is used for fitting the training data set, optimizing iteration parameter values and generating an output voltage limiting interval, and the specific steps comprise:
the gating cycle unit layer extracts the time sequence characteristics of the training data set;
the activation unit layer reads the training data set to generate the output voltage limit interval;
and the full connection layer carries out iterative computation based on the time sequence characteristics, the output voltage limit interval and the SOC data to generate an optimized iterative parameter value.
8. The all-vanadium redox flow battery voltage prediction method according to claim 7, wherein the all-connection layer comprises an error layer and a correction layer, wherein:
the error layer calculates a fitting error based on a loss function, the acquired time sequence characteristics, the output voltage limit interval and SOC data;
and if the fitting error is greater than or equal to a preset error threshold value, the correction layer generates the optimized iteration parameter.
9. The voltage prediction device for the all-vanadium redox flow battery is characterized in that: the all-vanadium redox flow battery voltage prediction device (100) comprises:
data acquisition module (10): the system is used for acquiring experimental data of the all-vanadium redox flow battery acquired by a monitoring system, and the all-vanadium redox flow battery is connected with an upper computer through the monitoring module;
data processing module (20): the system comprises a data acquisition module, a data storage module and a data processing module, wherein the data acquisition module is connected with the data acquisition module and is used for calculating the SOC of the all-vanadium redox flow battery under different working conditions according to the experimental data and preprocessing the experimental data and the corresponding SOC data to form a training data set;
and the neural network module (30) is connected with the data processing module and used for performing parameter fitting calculation by adopting a gated cyclic neural network model based on the training data set so as to predict the voltage value of the all-vanadium redox flow battery.
10. A computer-readable storage medium, in 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 claims 1 to 8.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210921665.0A CN115308608A (en) | 2022-08-02 | 2022-08-02 | All-vanadium redox flow battery voltage prediction method, device and medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202210921665.0A CN115308608A (en) | 2022-08-02 | 2022-08-02 | All-vanadium redox flow battery voltage prediction method, device and medium |
Publications (1)
Publication Number | Publication Date |
---|---|
CN115308608A true CN115308608A (en) | 2022-11-08 |
Family
ID=83858522
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202210921665.0A Pending CN115308608A (en) | 2022-08-02 | 2022-08-02 | All-vanadium redox flow battery voltage prediction method, device and medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN115308608A (en) |
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115566236A (en) * | 2022-12-05 | 2023-01-03 | 广东电网有限责任公司江门供电局 | Battery energy storage system operation control method, device, equipment and medium |
CN117007978A (en) * | 2023-10-07 | 2023-11-07 | 中国华能集团清洁能源技术研究院有限公司 | Battery voltage prediction method and device, electronic equipment and storage medium |
-
2022
- 2022-08-02 CN CN202210921665.0A patent/CN115308608A/en active Pending
Cited By (4)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN115566236A (en) * | 2022-12-05 | 2023-01-03 | 广东电网有限责任公司江门供电局 | Battery energy storage system operation control method, device, equipment and medium |
CN115566236B (en) * | 2022-12-05 | 2023-03-24 | 广东电网有限责任公司江门供电局 | Battery energy storage system operation control method, device, equipment and medium |
CN117007978A (en) * | 2023-10-07 | 2023-11-07 | 中国华能集团清洁能源技术研究院有限公司 | Battery voltage prediction method and device, electronic equipment and storage medium |
CN117007978B (en) * | 2023-10-07 | 2024-01-30 | 中国华能集团清洁能源技术研究院有限公司 | Battery voltage prediction method and device, electronic equipment and storage medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
Zhao et al. | Elman neural network using ant colony optimization algorithm for estimating of state of charge of lithium-ion battery | |
CN110187290B (en) | Lithium ion battery residual life prediction method based on fusion algorithm | |
CN108805217A (en) | A kind of health state of lithium ion battery method of estimation and system based on support vector machines | |
CN111832220A (en) | Lithium ion battery health state estimation method based on codec model | |
CN115308608A (en) | All-vanadium redox flow battery voltage prediction method, device and medium | |
CN113917334B (en) | Battery health state estimation method based on evolution LSTM self-encoder | |
CN115201686B (en) | Lithium ion battery health state assessment method under incomplete charge and discharge data | |
CN113608126B (en) | Lithium battery SOC online prediction method under different temperatures | |
CN113359048A (en) | Indirect prediction method for remaining service life of lithium ion battery | |
CN114839538A (en) | Method for extracting degradation characteristics of lithium ion battery for estimating residual life | |
CN114966436A (en) | Lithium battery state of charge prediction method, device, equipment and readable storage medium | |
Li et al. | A hybrid framework for predicting the remaining useful life of battery using Gaussian process regression | |
CN113933714A (en) | Battery capacity prediction method based on combination of simplified electrochemical model and grey prediction | |
CN115389946A (en) | Lithium battery health state estimation method based on isobaric rise energy and improved GRU | |
CN115792677A (en) | Lithium ion battery life prediction method based on improved ELM | |
Li et al. | Degradation prediction of proton exchange membrane fuel cell based on the multi-inputs Bi-directional long short-term memory | |
CN113093014B (en) | Online collaborative estimation method and system for SOH and SOC based on impedance parameters | |
CN114839542A (en) | Sample data set generation method and SOC estimation method for power lithium battery | |
Huang et al. | State of health estimation of lithium-ion batteries based on fine-tuning or rebuilding transfer learning strategies combined with new features mining | |
CN117825965A (en) | State prediction method and system for lithium ion battery | |
Kuang et al. | State-of-charge estimation hybrid method for lithium-ion batteries using BiGRU and AM co-modified Seq2Seq network and H-infinity filter | |
CN116400246B (en) | ACO-GMR-based energy storage power station battery health state estimation method | |
Xu et al. | A hybrid method for lithium-ion batteries state-of-charge estimation based on gated recurrent unit neural network and an adaptive unscented Kalman filter | |
CN117169743A (en) | Battery health state estimation method and device based on partial data and model fusion | |
CN116774045A (en) | Lithium battery health state prediction method based on HHO-SVR |
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 |