CN103633351A - Method for establishing temperature control strategy for fuel battery - Google Patents
Method for establishing temperature control strategy for fuel battery Download PDFInfo
- Publication number
- CN103633351A CN103633351A CN201310575173.1A CN201310575173A CN103633351A CN 103633351 A CN103633351 A CN 103633351A CN 201310575173 A CN201310575173 A CN 201310575173A CN 103633351 A CN103633351 A CN 103633351A
- Authority
- CN
- China
- Prior art keywords
- parameter
- control strategy
- temperature control
- fuel battery
- relation
- 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
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01M—PROCESSES OR MEANS, e.g. BATTERIES, FOR THE DIRECT CONVERSION OF CHEMICAL ENERGY INTO ELECTRICAL ENERGY
- H01M8/00—Fuel cells; Manufacture thereof
- H01M8/04—Auxiliary arrangements, e.g. for control of pressure or for circulation of fluids
- H01M8/04298—Processes for controlling fuel cells or fuel cell systems
- H01M8/04992—Processes for controlling fuel cells or fuel cell systems characterised by the implementation of mathematical or computational algorithms, e.g. feedback control loops, fuzzy logic, neural networks or artificial intelligence
-
- H—ELECTRICITY
- H01—ELECTRIC ELEMENTS
- H01M—PROCESSES OR MEANS, e.g. BATTERIES, FOR THE DIRECT CONVERSION OF CHEMICAL ENERGY INTO ELECTRICAL ENERGY
- H01M8/00—Fuel cells; Manufacture thereof
- H01M8/04—Auxiliary arrangements, e.g. for control of pressure or for circulation of fluids
- H01M8/04007—Auxiliary arrangements, e.g. for control of pressure or for circulation of fluids related to heat exchange
-
- 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
- Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- General Chemical & Material Sciences (AREA)
- Electrochemistry (AREA)
- Chemical Kinetics & Catalysis (AREA)
- Chemical & Material Sciences (AREA)
- Sustainable Energy (AREA)
- Sustainable Development (AREA)
- Manufacturing & Machinery (AREA)
- Computing Systems (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Medical Informatics (AREA)
- Fuzzy Systems (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Automation & Control Theory (AREA)
- Artificial Intelligence (AREA)
- Fuel Cell (AREA)
Abstract
The invention discloses a method for establish a temperature control strategy for a fuel battery. According to the method, originally collected data information samples of actual running parameters relevant to the fuel battery are analyzed through a support vector machine model by using a clustering analysis method and a correlation analysis method to obtain a relation between every two parameters; and therefore, the parameters and the parameter relations are subjected to machine learning, the parameter relation for establishing the temperature control strategy for the fuel battery is extracted, and a user can establish the temperature control strategy.
Description
Technical field
The present invention relates to a kind of formulating method of temperature intelligent control strategy of fuel cell.
Background technology
Fuel cell is become by air supply unit, pile, thermal management unit, DC/DC unit and other enclosure group, as shown in Figure 1, fuel cell is " generator " of back reaction that utilizes the electrolysis of water, during work, by hydrogen cylinder and air-in, fuel (hydrogen) and oxidant (oxygen) are input in pile jointly, by the combustion reaction of fuel, chemical energy is converted into electric energy, then by DC/DC module, voltage transitions is become in the electric pressure of the load that need to power.Fuel cell has the advantages such as generating efficiency is high, low in the pollution of the environment.Fuel cell applications is extensive, both can be applicable to military affairs, space, field, power plant, also can be applicable to the fields such as motor vehicle, mobile device, resident family.But fuel cell is very complicated, relate to numerous subject correlation theories such as chemical thermodynamics, electrochemistry, electro-catalysis, material science, electric power system and automatic control, how to improve the utilance of fuel cell, the reliability that how to guarantee fuel cell, fail safe etc. and be determine at present fuel cell whether can large-scale application in the subject matter of the people's livelihood, this wherein, the reasonable control of stack temperature is particularly important for the normal operation that ensures fuel cell.
From automation field direction, for the key job that improves the serviceability of fuel cell and extend its useful life, be just to provide rational control strategy.As everyone knows, the formulation of control strategy depends on determining of parameter and extracts, and comprises and determines which parameter is to determine controlling for temperature in certain purposes, certain pile, some region, some weather conditions, some policy conditions, certain user, some occasion or some equipment classification situation is between parameter and these parameters, how to combine or transmit the most effectively, reliably or association just can reach and controls stack temperature within normal range (NR).
Chinese patent 201110264545 " user controls the method for the temperature of fuel cell system " has been described the method for controlling the temperature of fuel cell system, comprise based on cooling liquid outlet temperature and cooling wind speed etc., but this method is only controlled stack temperature for limited parameter and is described the impact of not considering other factor temperature, its use face is also very narrow, also very limited to formulating reasonably comprehensive control strategy.
Chinese patent " 201110346837.8 " " a kind of formulating method of intelligent power consumption strategy " has been described the policy development method in a kind of intelligent power field, the method is passed through linear partition, extract neighbour, cluster, thereby the associated formulating method that waits relationship characteristic to obtain formulating control strategy, the method does not have for specific demand for control when extracting the relationship characteristics such as association, such as power requirement, battery utilance, loading demand etc., thereby and the method is not carried out machine learning again and is verified and can not guarantee the tactful reasonability of formulating from precision after extracting these relationship characteristics.
Summary of the invention
The object of the invention is to overcome the shortcoming of prior art, propose a kind of formulating method of fuel battery temperature control strategy.The present invention meets the every specified export control policy of fuel cell for formulating, and to guarantee fuel cell fail safe, reliability and durability in use, and provides the strategy support of wisdom for user and manager.
On the basis of the actual operation parameters relevant to fuel cell that the present invention receives on long-range unattended operation equipment, adopt the method for cluster analysis, association analysis, obtain the relation between parameter and parameter, on this basis these parameters and parameters relationship are carried out to machine learning, analysis obtains for formulating the parameter of fuel battery temperature control strategy, and the dependence syntagmatic between parameter, user can formulate temperature control strategy according to these analysis results.Such as, in spring, ambient temperature is between 15-20 degree Celsius time, if output voltage is greater than some values, although the pressure of pile is when normal range (NR) is still greater than some force value, the temperature of pile is just in critical point, in order rationally to control stack temperature, under this environmental condition, with regard to needing, according to the pressure data detecting, pile hydrogen cylinder pressure is regulated, formulate accordingly temperature control strategy.
The formulating method of fuel cell Intelligent Control Strategy of the present invention comprises following concrete steps:
1. on long-range unattended operation equipment, as Control card, receive the data message sample of Fuel Cell Control actual operation parameters, control parameter basic data, with the tight associated control parameter information such as temperature, safety, efficiency, such as data messages such as temperature, pressure, voltage, electric current, density of hydrogen, air mass flows;
2. adopt supporting vector machine model to carry out linear partition to sample space, the result after division forms each dimension in message sample space, comprises and controls parameter peacekeeping effectiveness dimension.Described control information dimension comprises the electrical nature such as electric weight dimension, electric current dimension, voltage dimension, active power dimension, reactive power dimension.Described effectiveness dimension comprises the control parameter attributes such as time dimension, Wei, region, place dimension, weather dimension, illuminance dimension, temperature dimension, humidity dimension, user type dimension, user gradation dimension, device levels dimension, device class dimension, operating space type dimension, operating space rank dimension, frequency of operation dimension, valid function frequency dimension, energy-conservation degree dimension, energy-conservation rank dimension;
3. in the sample space after linear partition, further find cluster feature, the linked character between each dimensional feature in above-mentioned steps 2, be specially:
1) the formulation cluster feature of dividing based on figure, the described clustering method of dividing based on figure, comprise that the figure based on boolean's link divides and the figure division based on weight link, and be divided into respectively the subgraphs of different sizes, to the node in specific subgraph, the factor of influence that is each dimensional feature in step 2 calculates arithmetic average, in order to generate the cluster relationship characteristic between each dimension in step 2;
2) the incidence relation analysis based on directed graph, adopts confidence level transmission, the confidence level transmission based on converse digraph and the confidence level transmission based on non-directed graph of directed graph to the parameter attribute described in each, and each dimensional feature in step 2, generates incidence relation feature;
3) on the parameter space after expansion, training set and test set are represented again, use PRELIMINARY RESULTS and second extraction result to represent sample simultaneously; Training classifier on training set, after grader is finished by training, uses the grader training to carry out policy learning to the website sample in test set, completes the optimization to primary learning.
4. relationship characteristic and combination thereof between the Fuel Cell Control parameter obtaining by above step, control object according to the temperature under different condition and carry out the formulation of temperature control strategy.
Accompanying drawing explanation
The fuel cell schematic diagram that Fig. 1 the present invention relates to;
The formulating method flow chart of Fig. 2 Fuel Cell Control method of the present invention.
Embodiment
Below in conjunction with the drawings and specific embodiments, the present invention is further described.
The overall procedure of fuel cell control strategy formulating method of the present invention is as shown in Figure 1: step S1 is preliminary tagsort, be that the Fuel Cell Control information that preliminary treatment is returned from long-range unattended operation equipment (as Control card) collection comprises the information such as temperature, pressure, voltage, electric current, density of hydrogen, air mass flow, it is carried out to dimension division.Step S2 is on the basis of preliminary tagsort result, carry out the extraction of further feature, extract cluster feature, linked character, in this course, factor of influence based on characteristic pattern and primary learning is extracted to this two category feature, this two category feature is carried out respectively obtaining final parameter and various effective combined result thereof after machine learning, thereby further support science, formulation and the enforcement of temperature control strategy effectively.As shown in Figure 2, first the control information data of collecting are carried out to linear partition, according to different dimensions, divide, such as being divided into time dimension, ambient temperature dimension, stack temperature dimension, enter hydrogen cylinder pressure dimension, go out hydrogen cylinder pressure dimension, load voltage dimension, load current dimension etc., then according to the clustering method that adopts non-directed graph, according to different principle of classification, by controlling parameter attribute network, be divided into K class with the cluster relation between analysis and Control parameter, the incidence relation between the controlled parameter of analysis of the support based on item collection and frequent item set simultaneously, then on the parameter attribute space after the expansion obtaining, training set and test set are represented again, use the new character representation sample extracting after preliminary feature and analysis simultaneously, training classifier on training set, the selection of grader can be any present mode grader, after grader is finished by training, uses the grader training to carry out policy learning to the sample in test set, completes the optimization to primary learning.
Specific as follows:
1, all Fuel Cell Control parameter informations in the current database of preliminary treatment, carry out the dimension classification of feature.
2,, on the basis of classification results, carry out the extraction of cluster feature, linked character.
On the basis of preliminary classification result, different parameters is extracted respectively to two classes relation of different nature, be cluster relation and incidence relation, adopt again machine learning algorithm to learn again used feature, finally obtain impact and formulate the parameter of temperature control strategy and the dependence syntagmatic between these parameters.
Extracting method with regard to cluster feature and linked character is described respectively below.
The extraction of described cluster feature is the feature of dividing based on figure, the clustering method that should divide based on figure, consider that the figure partitioning algorithm of existing maturation is mostly for non-directed graph, for simplified operation, all control parameter attribute networks relevant to fuel cell are regarded as to non-directed graph here and process simultaneously.
The extraction of described linked character, confidence level transmission, the confidence level transmission based on converse digraph and the confidence level transmission based on non-directed graph based on directed graph, each is controlled to parameter, generate transfer characteristic, the relevant control parameter attribute network of whole fuel cell is regarded as to directed graph here or non-directed graph is processed.
By the leaching process of the above cluster relationship characteristic and linked character, by the fuel cell described in step 2 respectively tie up on time between parameter and space can affect temperature controlled parameter and relation is found out, and can As time goes on constantly find to affect the new parameter of controlling effectiveness, extract parameter and parameters relationship combination that Bearing performance form is many bunches of multiple-limbs, can require to carry out different combinations and sequence according to different control performances, the management end of fuel cell or user side all can be suitable for optimum temperature control strategy instantly according to extracting result formulation.
Take below the formulation of fuel cell for communication control strategy as example illustrates step of the present invention.
Primary data sample is being carried out after linear classification, suppose to need extraction property is " in ambient temperature, being in the region situation between subzero 5 ° to above freezing 25; guarantee that pile power output is 3kw; keep the operating characteristics set of the constant every control parameter 50 ° of left and right of stack temperature " simultaneously, based on above-mentioned analytical method, the method that analysis obtains available strategy is as follows:
(1) adopt the clustering method of non-directed graph, according to different principle of classification, controls parameter attribute network is divided into K class, such as classifying according to temperature range, according to construction quality property sort, classify etc. according to device characteristics.This example adopt according to temperature range classification, according to output characteristic classification and according to controlling parameter grade classification and according to methods such as control operation classification.The cluster feature of calculating the special parameter in a cluster adopts following formula:
Wherein, E be all temperature or other data objects square error and, p represents the given object of data centralization, m
ia bunch C
icenter, each object represents initial average or the center of a bunch of Clustering, p and m
ican be multidimensional.That is, ask each object in each bunch to the quadratic sum of each bunch of centre distance.This criterion makes individual bunch of the k generating compact as much as possible and independent.
(2) analysis of the support based on item collection and frequent item set obtains the incidence relation between data.The support of its middle term collection: the support of a collection A is the percentage that comprises A in D, that is:
3,, on the basis of above analysis result, adopt machine learning algorithm to learn again strategy, and generate design result.
Described in step S4, adopting machine learning algorithm to learn used feature to strategy, is to be formed by the feature of the preliminary policy learning of step S1 and the parameter combinations of step S2 formulation again.
Described strategy is learnt specifically to comprise again: on the feature space after expansion, training set and test set are represented again, use preliminary feature and second extraction character representation sample simultaneously; Training classifier on training set, the selection of grader can be any present mode grader, after grader is finished by training, uses the grader training to carry out policy learning to the website sample in test set, complete the optimization to primary learning, generate final Strategy Design result.
By above method, can obtain impact and formulate the various set of temperature controlled parameters relationship, thereby generation control strategy, object factory is the operative combination that satisfies the demands and controls the relation between parameter, such as the increase of hydrogen cylinder pressure or reduce, ventilating fan rotating speed, the control of air inlet temperature be, the control of control, air mass flow and the density of hydrogen of counterbalance valve pressure.
Claims (5)
1. the formulating method of a fuel battery temperature control strategy, it is characterized in that, on the basis of the data message sample of the actual operation parameters relevant to fuel cell that described method obtains in original collection, adopt the method for supporting vector machine model and cluster analysis, association analysis to analyze, obtain the relation between parameter and parameter, on this basis these parameters and parameters relationship are carried out to machine learning, extract for formulating the parameters relationship of fuel battery temperature control strategy, for user, formulate temperature control strategy.
2. according to the formulating method of fuel battery temperature control strategy claimed in claim 1, it is characterized in that, described extraction comprises the following steps for formulating the method for fuel battery temperature control strategy parameters relationship:
1) from long-range unattended operation equipment, collect the message sample of fuel cell relevant parameter;
2) adopt supporting vector machine model to carry out linear partition to sample space, the result formation control parameter peacekeeping effectiveness dimension of linear partition;
3) in the sample space after division according to each parameter the factor of influence to temperature parameter, further find between each relevant parameter of incidence relation and cluster relation to(for) temperature;
4), on the basis of the result of above searching relation, adopt machine learning algorithm to learn again parameter and parameters relationship;
5) obtain parameter and the parameters relationship that fuel battery temperature control strategy is formulated in final impact.
3. the formulating method of fuel cell control strategy according to claim 2, is characterized in that, the cluster relation of described step 3) is based on K-Means Algorithm Analysis cluster feature.
4. the formulating method of fuel battery temperature control strategy according to claim 2, is characterized in that, the incidence relation of described step 3) is to extract based on frequent item set, associated confidence level.
5. the formulating method of fuel battery temperature control strategy according to claim 2, the machine learning algorithm adopting in described step 4) specifically comprises:
On parameter space after study, training set and test set are represented again, control parameter association relation and the cluster relation of using described step 3) to extract represent sample simultaneously; Training classifier on training set, after grader training finishes, uses the grader training to learn the sample in test set again, completes the optimization to primary learning.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310575173.1A CN103633351A (en) | 2013-11-15 | 2013-11-15 | Method for establishing temperature control strategy for fuel battery |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201310575173.1A CN103633351A (en) | 2013-11-15 | 2013-11-15 | Method for establishing temperature control strategy for fuel battery |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN103633351A true CN103633351A (en) | 2014-03-12 |
Family
ID=50214136
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201310575173.1A Pending CN103633351A (en) | 2013-11-15 | 2013-11-15 | Method for establishing temperature control strategy for fuel battery |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN103633351A (en) |
Cited By (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104167570A (en) * | 2014-05-30 | 2014-11-26 | 杭州电子科技大学 | Rapid assembling method for storage battery |
| CN110823291A (en) * | 2019-11-27 | 2020-02-21 | 山东建筑大学 | Method and system for monitoring indoor temperature and humidity environment of building based on K-means clustering algorithm |
| CN110867597A (en) * | 2019-11-21 | 2020-03-06 | 电子科技大学 | Thermoelectric water cooperative control method for consistency of proton exchange membrane fuel cell |
| CN112201822A (en) * | 2020-09-16 | 2021-01-08 | 武汉海亿新能源科技有限公司 | Temperature self-learning cooling method, device and system for hydrogen fuel cell |
| CN116150566A (en) * | 2023-04-20 | 2023-05-23 | 浙江浙能迈领环境科技有限公司 | Ship fuel supply safety monitoring system and method thereof |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102426676A (en) * | 2011-11-06 | 2012-04-25 | 中国科学院电工研究所 | Feature extraction method of intelligent power consumption strategy |
| CN103018673A (en) * | 2012-11-19 | 2013-04-03 | 北京航空航天大学 | Method for predicating life of aerospace Ni-Cd storage battery based on improved dynamic wavelet neural network |
-
2013
- 2013-11-15 CN CN201310575173.1A patent/CN103633351A/en active Pending
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN102426676A (en) * | 2011-11-06 | 2012-04-25 | 中国科学院电工研究所 | Feature extraction method of intelligent power consumption strategy |
| CN103018673A (en) * | 2012-11-19 | 2013-04-03 | 北京航空航天大学 | Method for predicating life of aerospace Ni-Cd storage battery based on improved dynamic wavelet neural network |
Cited By (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104167570A (en) * | 2014-05-30 | 2014-11-26 | 杭州电子科技大学 | Rapid assembling method for storage battery |
| CN110867597A (en) * | 2019-11-21 | 2020-03-06 | 电子科技大学 | Thermoelectric water cooperative control method for consistency of proton exchange membrane fuel cell |
| CN110867597B (en) * | 2019-11-21 | 2022-06-14 | 电子科技大学 | Thermoelectric water cooperative control method for consistency of proton exchange membrane fuel cell |
| CN110823291A (en) * | 2019-11-27 | 2020-02-21 | 山东建筑大学 | Method and system for monitoring indoor temperature and humidity environment of building based on K-means clustering algorithm |
| CN112201822A (en) * | 2020-09-16 | 2021-01-08 | 武汉海亿新能源科技有限公司 | Temperature self-learning cooling method, device and system for hydrogen fuel cell |
| CN116150566A (en) * | 2023-04-20 | 2023-05-23 | 浙江浙能迈领环境科技有限公司 | Ship fuel supply safety monitoring system and method thereof |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN103633351A (en) | Method for establishing temperature control strategy for fuel battery | |
| CN105069519A (en) | Intelligent power grid park terminal user energy demand condition dynamic prediction system and method | |
| CN104951834A (en) | LSSVM (least squares support vector machine) wind speed forecasting method based on integration of GA (genetic algorithm) and PSO (particle swarm optimization) | |
| CN104881735A (en) | System and method of smart power grid big data mining for supporting smart city operation management | |
| CN113780684A (en) | Intelligent building user energy consumption behavior prediction method based on LSTM neural network | |
| Zhang et al. | An optimal model identification for solid oxide fuel cell based on extreme learning machines optimized by improved Red Fox Optimization algorithm | |
| CN103400039A (en) | Wind power climbing prediction model switching method based on gale weather classification | |
| CN106294738A (en) | A kind of Intelligent household scene collocation method | |
| CN106296315A (en) | Context aware systems based on user power utilization data | |
| Liang et al. | Evaluation of battery modules state for electric vehicle using artificial neural network and experimental validation | |
| CN115563885A (en) | Energy conversion system for energy storage battery of wind power plant | |
| Li et al. | Residential Electricity Classification Method Based On Cloud Computing Platform and Random Forest. | |
| Yang | Research on integration method of AI teaching resources based on learning behaviour data analysis | |
| Gamboa-Rosales et al. | Determining Technologies Trends and Evolution of Smart Building Technologies by Bibliometric Analysis from 1984 to 2020: Trends, research opportunities and future perspectives | |
| Olu-Ajayi et al. | Ensemble learning for energy performance prediction of residential buildings | |
| CN109378834A (en) | Large scale electric network voltage stability margin assessment system based on information maximal correlation | |
| CN108920790A (en) | A kind of public building electricity consumption pattern recognition model method for building up based on historical data | |
| CN116777153A (en) | Distribution network flexibility attribution analysis method considering distributed energy access | |
| Liu et al. | An Efficient Supervised Energy Disaggregation Scheme for Power Service in Smart Grid. | |
| CN115313377A (en) | Power load prediction method and system | |
| Qin | Software reliability prediction model based on PSO and SVM | |
| CN111768066B (en) | Park electric heating load coupling relation analysis method and device based on fusion characteristics | |
| Lu et al. | Anomaly Recognition Method for Massive Data of Power Internet of Things Based on Bayesian Belief Network | |
| Fan et al. | Fault diagnosis for fuel cell based on naive Bayesian classification | |
| Zhang | A method for monitoring the spatial distribution of nutrients from the perspective of landscape ecology |
Legal Events
| Date | Code | Title | Description |
|---|---|---|---|
| PB01 | Publication | ||
| PB01 | Publication | ||
| C10 | Entry into substantive examination | ||
| SE01 | Entry into force of request for substantive examination | ||
| C02 | Deemed withdrawal of patent application after publication (patent law 2001) | ||
| WD01 | Invention patent application deemed withdrawn after publication |
Application publication date: 20140312 |