CN107679649A - A kind of failure prediction method of electrical equipment, device, storage medium and electrical equipment - Google Patents
A kind of failure prediction method of electrical equipment, device, storage medium and electrical equipment Download PDFInfo
- Publication number
- CN107679649A CN107679649A CN201710824151.2A CN201710824151A CN107679649A CN 107679649 A CN107679649 A CN 107679649A CN 201710824151 A CN201710824151 A CN 201710824151A CN 107679649 A CN107679649 A CN 107679649A
- Authority
- CN
- China
- Prior art keywords
- electrical equipment
- sample data
- history
- training
- service condition
- 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
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- 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
- G06N3/084—Backpropagation, e.g. using gradient descent
Landscapes
- Engineering & Computer Science (AREA)
- Business, Economics & Management (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Strategic Management (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- Computational Linguistics (AREA)
- Development Economics (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Data Mining & Analysis (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Biophysics (AREA)
- Biomedical Technology (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Game Theory and Decision Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Entrepreneurship & Innovation (AREA)
- Marketing (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention discloses a kind of failure prediction method of electrical equipment, device, storage medium and electrical equipment, this method includes:The relation established between the history run state of electrical equipment and historical failure situation;Obtain the actual motion state for the electrical equipment for having established the relation;By the historical failure situation corresponding to the history run state corresponding with the actual motion state in the relation, as physical fault situation corresponding with the actual motion state, to obtain the failure predication result to electrical equipment.The solution of the present invention, can overcome unit is damaged in the prior art it is big, safeguard the defects of promptness difference and poor user experience, realize unit is damaged it is small, safeguard the beneficial effect that promptness is good and Consumer's Experience is good.
Description
Technical field
The invention belongs to air-conditioning technical field, and in particular to a kind of failure prediction method of electrical equipment, device, storage medium and
Electrical equipment, more particularly to a kind of method of air-conditioning failure predication, device corresponding with this method, the calculating for being stored with this method instruction
Machine readable storage medium storing program for executing and it is able to carry out this method instruction or is stored with the electrical equipment of the device.
Background technology
Air-conditioning (i.e. air regulator), can be to temperature, humidity, cleanliness factor, the speed of surrounding air in building/structures
It is adjusted and controls etc. parameter.For air-conditioning when user uses process normal operation, operational factor deviates normal operation shape at present
State, until when air-conditioning has been subjected to the inspection logic judgment of itself and is out of order, now air-conditioning can not normal use, it is necessary to which user enters
Row reports for repairment, professional carries out maintenance operation, on the one hand damages larger, on the one hand influence user use in itself to air-conditioner set.
In the prior art, exist it is big to unit infringement, safeguard the defects of promptness difference and poor user experience.
The content of the invention
It is an object of the present invention to it is directed to drawbacks described above, there is provided a kind of failure prediction method of electrical equipment, device, storage are situated between
Matter and electrical equipment, with solve air-conditioning is out of order by the inspection logic judgment of itself in the prior art when can not normal use cause
The problem of big to unit infringement, reach and small effect is damaged to unit.
The present invention provides a kind of failure prediction method of electrical equipment, including:Establish history run state and the history event of electrical equipment
Relation between barrier situation;Obtain the actual motion state for the electrical equipment for having established the relation;By in the relation with the reality
The border running status historical failure situation corresponding to the history run state accordingly, as with the actual motion state
Corresponding physical fault situation, to obtain the failure predication result to electrical equipment.
Alternatively, wherein, the relation established between the history run state of electrical equipment and historical failure situation, including:Collect
History service condition of the electrical equipment of same type in different user;Wherein, the history service condition, including:Electric operation
During history run state and historical failure situation;By big data analysis and digging technology, feelings are used to the history
Condition is analyzed, and is obtained using the history run state as input parameter and using the historical failure situation as output parameter
Data pair, as sample data;Using neural network algorithm, the sample data is trained and tested, needed for obtaining
The history run state and the historical failure situation between relation.
Alternatively, wherein, history service condition of the electrical equipment of same type in different user is collected, including:Pass through nothing
Line R-T unit, receive history service condition of the electrical equipment of same type in preset duration in different user;And/or to institute
History service condition is stated to be analyzed, including:According at least one fault type of required prediction, SVMs is utilized
At least one of habit, machine learning, data analysis mode, the history service condition is analyzed, obtains analysis result;
With reference to default expertise, will have an impact in the analysis result to the fault type, and/or have with the fault type
The history run state of association is as input parameter, and by historical failure corresponding with the input parameter in the analysis result
Situation is as output parameter;And/or the sample data is trained and tested, including:The sample data is divided into
Training sample data and test sample data;The network structure of neural network algorithm is chosen, initializes the company of the network structure
Connect weights;By the network structure and the connection weight, the training sample data are trained;Pass through the training
The network structure and connection weight of completion, the test sample data are tested.
Alternatively, wherein, the training sample data are trained, including:By the input of the training sample data
Parameter is input in the network structure, obtains theoretical output parameter;Obtain the theoretical output parameter and the training sample
Training error between the output parameter of data, and determine the training error whether in the range of target training error;Work as institute
When stating training error in the range of the target training error, deconditioning;Or when the training error is trained in the target
When outside error range, the connection weight is adjusted;And/or the test sample data are tested, including:Will
The input parameter of the test sample data is inputted in the network structure, obtains theoretical output parameter;It is defeated to obtain the theory
The test error gone out between the output parameter of parameter and the test sample data, and determine the test error whether in target
In the range of test error;When the test error is in the target detection error range, stop test;Or when the test
When error is outside the target detection error range, network structure described in re -training;And/or expand the electrical equipment of same type
The capture range of history service condition in different user, used with history of the electrical equipment to same type in different user
Situation is collected again.
Alternatively, wherein, in the history service condition, the history run state and the historical failure situation,
It is sequentially arranged;And/or the input parameter, including:Single history run state;And/or by setting rule
Individual features are extracted from the history run state, and the one-dimension array or bidimensional above array being made up of the feature;With/
Or, the network structure, including:At least one of input node number, output node number, network number of plies;And/or to described
Connection weight is adjusted, including:According to the adjustment system related to the time series of the connection weight and the training sample
Number, is adjusted by training error back transfer method to the connection weight.
Alternatively, in addition to:According to the failure predication result, determine the physical fault situation fault degree whether
Reach the early warning degree of setting;When the fault degree of the physical fault situation reaches the early warning degree of setting, prompting is initiated;
And/or the relation between the history run state and the historical failure situation is updated;And/or to the history
At least one of running status, the historical failure situation, the actual motion state, described physical fault situation are shown
Show.
Alternatively, wherein, the relation between the history run state and the historical failure situation is updated, wrapped
Include:When also the connection weight including pair network structure related to the relation is adjusted this method, same type is collected
New service condition of the electrical equipment in different user;Using new service condition as new sample data, to the network knot
The connection weight of structure is adjusted;And/or by new service condition and history service condition together as new sample data,
To for adjusting after the regulation coefficient of the connection weight is adjusted, network structure described in re -training;And/or the event
Hinder prediction result, including:Physical fault situation from current time after a period of time.
Matching with the above method, another aspect of the present invention provides a kind of fault prediction device of electrical equipment, including:Control is single
Member, for establishing the relation between the history run state of electrical equipment and historical failure situation;Communication unit, established for obtaining
The actual motion state of the electrical equipment of the relation;Described control unit, be additionally operable to by the relation with the actual motion shape
The state historical failure situation corresponding to the history run state accordingly, as corresponding with the actual motion state real
Border failure situation, to obtain the failure predication result to electrical equipment.
Alternatively, wherein, the pass that described control unit is established between the history run state of electrical equipment and historical failure situation
System, specifically includes:Collect history service condition of the electrical equipment of same type in different user;Wherein, the history uses feelings
Condition, including:History run state and historical failure situation during electric operation;By big data analysis and digging technology,
The history service condition is analyzed, obtained using the history run state as input parameter and with the historical failure
Situation is the data pair of output parameter, as sample data;Using neural network algorithm, the sample data is trained and
Test, to obtain the relation between the required history run state and the historical failure situation.
Alternatively, wherein, described control unit collects history service condition of the electrical equipment of same type in different user,
Specifically include:By wireless transmitter, receive history of the electrical equipment of same type in preset duration in different user and use
Situation;And/or described control unit is analyzed the history service condition, is specifically included:According to required prediction at least
A kind of fault type, learnt using SVMs, machine learning, at least one of data analysis mode, to the history
Service condition is analyzed, and obtains analysis result;With reference to default expertise, by the analysis result to the failure classes
Type has an impact, and/or with the related history run state of the fault type as input parameter, and by the analysis result
In historical failure situation corresponding with the input parameter as output parameter;And/or described control unit is to the sample number
According to being trained and testing, specifically include:The sample data is divided into training sample data and test sample data;Choose
The network structure of neural network algorithm, initialize the connection weight of the network structure;Pass through the network structure and the company
Weights are connect, the training sample data are trained;The network structure and connection weight completed by the training, to described
Test sample data are tested.
Alternatively, wherein, described control unit is trained to the training sample data, is specifically included:By the instruction
The input parameter for practicing sample data is input in the network structure, obtains theoretical output parameter;Obtain the theoretical output ginseng
Several training errors between the output parameter of the training sample data, and determine whether the training error trains in target
In error range;When the training error is in the range of the target training error, deconditioning;Or when the training error
When outside the target training error scope, the connection weight is adjusted;And/or described control unit is to the survey
Sample notebook data is tested, and is specifically included:The input parameter of the test sample data is inputted in the network structure, obtained
To theoretical output parameter;The test obtained between the theoretical output parameter and the output parameter of the test sample data misses
Difference, and determine the test error whether in target detection error range;When the test error misses in the target detection
When in poor scope, stop test;Or when the test error is outside the target detection error range, net described in re -training
Network structure;And/or expand the capture range of history service condition of the electrical equipment of same type in different user, with to identical
History service condition of the electrical equipment of type in different user is collected again.
Alternatively, wherein, in the history service condition, the history run state and the historical failure situation,
It is sequentially arranged;And/or the input parameter, including:Single history run state;And/or by setting rule
Individual features are extracted from the history run state, and the one-dimension array or bidimensional above array being made up of the feature;With/
Or, the network structure, including:At least one of input node number, output node number, network number of plies;And/or the control
Unit processed is adjusted to the connection weight, is specifically included:According to the time with the connection weight and the training sample
The related regulation coefficient of sequence, is adjusted by training error back transfer method to the connection weight.
Alternatively, in addition to:Described control unit, it is additionally operable to according to the failure predication result, determines the actual event
Whether the fault degree of barrier situation reaches the early warning degree of setting;When the fault degree of the physical fault situation reaches setting
During early warning degree, prompting is initiated;And/or described control unit, it is additionally operable to the history run state and the historical failure
Relation between situation is updated;And/or described control unit, it is additionally operable to the history run state, history event
At least one of barrier situation, the actual motion state, described physical fault situation are shown.
Alternatively, wherein, described control unit is to the pass between the history run state and the historical failure situation
System is updated, and specifically includes:When the device is also additionally operable to a pair network knot related to the relation including described control unit
When the connection weight of structure is adjusted, new service condition of the electrical equipment of same type in different user is collected;New is made
By the use of situation as new sample data, the connection weight of the network structure is adjusted;And/or by new service condition
With history service condition together as new sample data, the regulation coefficient for adjusting the connection weight is adjusted
Afterwards, network structure described in re -training;And/or the failure predication result, including:From current time after a period of time
Physical fault situation.
Matching with the above method, further aspect of the present invention provides a kind of storage medium, including:Deposited in the storage medium
Contain a plurality of instruction;The a plurality of instruction, for being loaded by processor and being performed the failure prediction method of above-described electrical equipment.
Matching with the above method or device, further aspect of the present invention provides a kind of air-conditioning, including:Processor, for holding
The a plurality of instruction of row;Memory, for storing a plurality of instruction;Wherein, a plurality of instruction, for by the memory storage, and
Loaded by the processor and perform the failure prediction method of above-described electrical equipment;Or the failure of above-described electrical equipment
Prediction meanss.
The solution of the present invention, by the generation of look-ahead air-conditioning failure, provided to carry out protection and maintenance to air-conditioning in time
Foundation, can lower the use of the infringement and the reduction of more lower bound degree of air-conditioner set on user influences.
Further, the solution of the present invention, entered by the substantial amounts of history run state of the air-conditioner set of collection and failure situation
Row analysis and excavation, establish the running status of air-conditioning and the relation of failure situation, right early for the generation of look-ahead failure
Air-conditioning carries out Inspection and maintenance, then can lower the use of the infringement and the reduction of more lower bound degree of air-conditioner set on user influences.
Further, the solution of the present invention, by before air-conditioning substantially breaks down with regard to user or after sale people can be prompted
Member's air-conditioner set may be broken down and point out fault type, and the infringement and more lower bound degree that can lower air-conditioner set are reduced to user
Use influence.
Further, the solution of the present invention, by the relation of the running status to air-conditioning and failure situation safeguard and more
Newly, the accuracy rate of prediction can be improved.
Thus, the solution of the present invention, by establishing the running status of air-conditioning and the relation of failure situation, according to the fortune of air-conditioning
The failure situation of row status predication air-conditioning, solve when air-conditioning is out of order by the inspection logic judgment of itself in the prior art without
Method normal use causes the problem of big to unit infringement, so as to, overcome unit is damaged in the prior art it is big, safeguard that promptness is poor
And the defects of poor user experience, realize to unit damage it is small, safeguard the beneficial effect that promptness is good and Consumer's Experience is good.
Other features and advantages of the present invention will be illustrated in the following description, also, partly becomes from specification
Obtain it is clear that or being understood by implementing the present invention.
Below by drawings and examples, technical scheme is described in further detail.
Brief description of the drawings
Fig. 1 is the schematic flow sheet of an embodiment of the failure prediction method of the electrical equipment of the present invention;
Fig. 2 is the one of the relation established in the method for the present invention between the history run state of electrical equipment and historical failure situation
The schematic flow sheet of embodiment;
Fig. 3 is the schematic flow sheet for the embodiment analyzed in the method for the present invention the history service condition;
Fig. 4 is the flow signal for the embodiment that the sample data is trained and tested in the method for the present invention
Figure;
Fig. 5 is the schematic flow sheet for the embodiment being trained in the method for the present invention to the training sample data;
Fig. 6 is the schematic flow sheet for the embodiment tested in the method for the present invention the test sample data;
Fig. 7 is the schematic flow sheet of an embodiment of fault pre-alarming in method of the invention;
Fig. 8 is the structural representation of an embodiment of the fault prediction device of the electrical equipment of the present invention;
Fig. 9 is the principle schematic of an embodiment of the fault prediction device of air-conditioning in electrical equipment of the invention.
With reference to accompanying drawing, reference is as follows in the embodiment of the present invention:
102- control units;104- communication units.
Embodiment
To make the object, technical solutions and advantages of the present invention clearer, below in conjunction with the specific embodiment of the invention and
Technical solution of the present invention is clearly and completely described corresponding accompanying drawing.Obviously, described embodiment is only the present invention one
Section Example, rather than whole embodiments.Based on the embodiment in the present invention, those of ordinary skill in the art are not doing
Go out under the premise of creative work the every other embodiment obtained, belong to the scope of protection of the invention.
According to an embodiment of the invention, there is provided a kind of method of failure prediction method of electrical equipment, the as shown in Figure 1 present invention
An embodiment schematic flow sheet.The failure prediction method of the electrical equipment can include:
At step S110, the relation established between the history run state of electrical equipment and historical failure situation.
In an optional example, the history run state of electrical equipment is established in method that can be of the invention with reference to shown in Fig. 2
The schematic flow sheet of one embodiment of the relation between historical failure situation, further illustrate in step S110 and establish electrical equipment
The detailed process of relation between history run state and historical failure situation.
Step S210, collect history service condition of the electrical equipment of same type in different user.
Wherein, the history service condition, can include:History run state and historical failure during electric operation
Situation.
Such as:By predicting the system and device of air-conditioning failure, service condition of the same class air-conditioning in different user is collected,
Including air-conditioning the various state parameters on running (containing operational factor, malfunction etc.).
Alternatively, in the history service condition, the history run state and the historical failure situation, on time
Between sequentially arrange.
Such as:Air-conditioning in the previous year in service data collect, and arranged in chronological order.According to time sequence
Arrange obtained malfunction.
Thus, by being sequentially arranged history run state and historical failure situation, be advantageous to lifting and closed to corresponding
It is the accuracy determined, also helps the reliability of lifting failure predication.
Alternatively, history service condition of the electrical equipment of same type in different user is collected in step S210, can be wrapped
Include:By wireless transmitter, history service condition of the electrical equipment of same type in preset duration in different user is received.
Such as:The service data being collected into, the running state of air conditioner of a period of time is referred to, and be more than some moment
The state of point.
Thus, by collecting the history service condition in a period of time, be advantageous to obtain mass data and then lifted to going through
The accuracy and reliability that history running status and historical failure situation determine.
Step S220, by big data analysis and digging technology, the history service condition is analyzed, obtained with institute
It is input parameter and the data pair using the historical failure situation as output parameter to state history run state, as sample data.
Wherein, the input parameter, can include:Single history run state;And/or as setting rule described in
Individual features are extracted in history run state, and the one-dimension array or bidimensional above array being made up of the feature.
Such as:Input parameter can be not only single parameter, also include the input for extracting feature composition according to certain rules
Parameter one or more dimensions array.
Thus, by the input parameter of diversified forms, be advantageous to be lifted flexibility and the convenience of input mode.
It is alternatively possible to one analyzed in method of the invention with reference to shown in Fig. 3 the history service condition is real
The schematic flow sheet of example is applied, further illustrates the detailed process analyzed in step S220 the history service condition.
Step S310, according at least one fault type of required prediction, learnt using SVMs, machine learning,
At least one of data analysis mode, the history service condition is analyzed, obtains analysis result.
Step S320, with reference to default expertise, will have an impact in the analysis result to the fault type and/
Or with the related history run state of the fault type as input parameter, and by the analysis result with the input
The corresponding historical failure situation of parameter is as output parameter.
Such as:Some failures for needing to predict are determined in advance, by the analysis to data and combine expertise knowledge,
Choose respectively on that may have influence or relevant state parameter on these failures as input parameter, corresponding event
Barrier state is as output parameter.
Thus, by being analyzed by required fault type history service condition, and expertise, Ke Yiti are combined
Rise the accuracy determined to relation between history run state and historical failure situation and reliability.
Step S230, using neural network algorithm, the sample data is trained and tested, to obtain required institute
State the relation between history run state and the historical failure situation.
Such as:It is substantial amounts of by the air-conditioner set to collection using related algorithmic techniques such as big data analysis and excavations
History run state and failure situation are analyzed and excavated, using such as SVMs or other machines learning method or
Data analysing method, the non-linear relation between running state of air conditioner parameter and failure or rule are excavated and learnt, so as to
Algorithm foundation is provided for real-time failure predication.
Such as:For existing input and output set, such as SVMs or other machines learning method can be passed through
Or the relation of the analysis such as data analysing method input and outlet chamber, it is assumed that y is output, and (x can be one-dimensional or more to x for input
The array of group), existence function relation between y and x:Y=f (x).
Wherein, many some known variables in function f be present, and such as SVMs or other machines learn or number
According to analysis scheduling algorithm, this corresponding relation can be gone out according to existing input and output approximate simulation, that is, solve some unknown changes in f
The value of amount, obtain definite functional expression f.
Thus, by collecting the history service condition of electrical equipment, analyzed and excavated, obtain history in history service condition
Corresponding relation between running status and historical failure situation, be advantageous to be lifted the accuracy and reliability of failure predication.
It is alternatively possible to the sample data is trained and tested in method of the invention with reference to shown in Fig. 4 one
The schematic flow sheet of embodiment, further illustrate the specific mistake that the sample data is trained and tested in step S230
Journey.
Step S410, the sample data is divided into training sample data and test sample data.
Such as:Using the part in the sample data as training sample data, another part is as test sample number
According to.
Such as:By obtained input, output parameter pair, a part is used as training book sample data, and a part is used as test specimens
Notebook data.
Step S420, the network structure of neural network algorithm is chosen, initialize the connection weight of the network structure.
Wherein, the network structure, can include:In input node number, output node number, the network number of plies at least it
One.
Thus, by the network structure of diversified forms, training for promotion and the flexibility tested and versatility are advantageous to.
Such as:Choose network:According to the input of selection, output parameter, select LSTM network structures (such as:The network knot
Structure, it can include:Input and output nodal point number, network number of plies etc.), initialize network connection weight wk。
Step S430, by the network structure and the connection weight, the training sample data are trained.
More it is alternatively possible to the training sample data are trained in method of the invention with reference to shown in Fig. 5 one
The schematic flow sheet of embodiment, further illustrate the detailed process being trained in step S430 to the training sample data.
Step S510, the input parameter of the training sample data is input in the network structure, obtained theoretical defeated
Go out parameter.
Step S520, the training obtained between the theoretical output parameter and the output parameter of the training sample data miss
Difference, and determine the training error whether in the range of target training error.
Step S530, when the training error is in the range of the target training error, deconditioning.
Or step S540, when the training error is outside the target training error scope, to the connection weight (example
Such as:wk) be adjusted.
Such as:Training network:The input quantity of training sample data is input in network, obtains theoretical output valve a (x),
Corresponding real output value y (x) is compared to obtain error amount e (x)=‖ y (x)-a (x) ‖ with sample data.
If e (x) meets within desired value, i.e., | e (x) |<During ∈, deconditioning;Otherwise, according to Feedback error etc.
Method is to wkIt is adjusted.
Thus, by being trained to training sample data, required history run state and historical failure feelings can be obtained
Preliminary relationship between condition, reliability is high, and accuracy is good.
In a more optional specific example, the connection weight is adjusted in step S540, can be included:According to
The regulation coefficient related to the time series of the connection weight and the training sample (such as:Parameter η), pass through training error
Back transfer method is adjusted to the connection weight.
Such as:According to the methods of Feedback error to wkIt is adjusted:
C (w, b) is the error energy function (by taking standard variance function as an example) of training set, and n is the total quantity of training sample,
Summation is carried out on total training sample x:
Update weights:
Parameter η value can be preset in formula 2, can also be adjusted according to time series (for example time-sequencing is rearward
Training data, η values are larger, then role is larger).
Thus, by being adjusted to connection weight, reliability of the lifting to training network structure is advantageous to, and then can be with
The accuracy of training for promotion result.
The test sample data are carried out by step S440, the network structure and connection weight completed by the training
Test.
Thus, by being trained to training sample data, then based on training result test sample data are tested,
It can obtain the corresponding relation of more accurate, relatively reliable history run state and historical failure situation, processing procedure is reliable,
Safety, processing mode are easy, accurate.
More it is alternatively possible to the test sample data are tested in method of the invention with reference to shown in Fig. 6 one
The schematic flow sheet of embodiment, further illustrate the detailed process tested in step S440 the test sample data.
Step S610, the input parameter of the test sample data is inputted in the network structure, obtains theoretical output
Parameter.
Step S620, the test obtained between the theoretical output parameter and the output parameter of the test sample data miss
Difference, and determine the test error whether in target detection error range.
Step S630, when the test error is in the target detection error range, stop test.
Or step S640, when the test error is outside the target detection error range, network described in re -training
Structure;And/or expand the capture range of history service condition of the electrical equipment of same type in different user, with to mutually similar
History service condition of the electrical equipment of type in different user is collected again.
Such as:The network that test sample data input has been trained, if the error energy function C of test set is unsatisfactory for
, it is necessary to which the step of repeating foregoing training network, if necessary, collects more data or the other input of selection again during preset value
Parameter.
Thus, by being tested on the basis of training result, to be verified to training result, lifting pair is advantageous to
The accuracy and reliability that relation between history run state and historical failure situation determines.
At step S120, the actual motion state for the electrical equipment for having established the relation is obtained.
It is at step S130, the history run state corresponding with the actual motion state in the relation is corresponding
The historical failure situation, as physical fault situation corresponding with the actual motion state, with obtain it is to electrical equipment therefore
Hinder prediction result.
Thus, by establishing the relation between history run state and historical failure situation, according to the relation according to reality
The failure that future may occur running status is predicted, with the generation of look-ahead air-conditioning failure, to enter in time to air-conditioning
Row protection and maintenance provides foundation, and can lower the use of the infringement and the reduction of more lower bound degree of air-conditioner set on user influences.
Wherein, the failure predication result, can include:Physical fault feelings from current time after a period of time
Condition.
Thus, by the prediction to following a period of time internal fault situation, electric operation situation can be grasped in advance, and then
Safeguarded in time when needing to safeguard, on the one hand can ensure electric operation safety, reliability height;On the other hand user can be ensured
Easy to use, hommization is good.
In an optional embodiment, it can also include:The process of fault pre-alarming.
It is alternatively possible in method of the invention with reference to shown in Fig. 7 an embodiment of fault pre-alarming schematic flow sheet, enter
One step illustrates the detailed process of fault pre-alarming.
Step S710, according to the failure predication result, determine whether the fault degree of the physical fault situation reaches
The early warning degree of setting.
Step S720, when the fault degree of the physical fault situation reaches the early warning degree of setting, prompting is initiated, with
User is prompted to safeguard or intervene in time, so as to the convenience for ensureing the security of electric operation and user uses.
Such as:Network has been trained, that is, has established y=f (x).Intelligent control center utilizes the network trained,
The data of the air-conditioning the past period arrived by the use of real-time collecting then obtain the failure shape of a period of time from now on as inputting
State, it will occur if prediction result is failure, information is delivered to terminal air-conditioning step by step, sending warning by air-conditioning reminds use
Report for repairment or ask for help in family.
Such as:Before air-conditioning substantially breaks down with regard to can prompt user or after sale personnel's air-conditioner set may occur therefore
Hinder and point out fault type.
Thus, by being prompted when failure needs to safeguard, to remind user to carry out Inspection and maintenance in time, and then lifted
The convenience that the security of electric operation and user use.
In an optional embodiment, it can also include:To the history run state and the historical failure situation
Between relation be updated.
Such as:The network algorithm structure of establishment, it is not unalterable, but can be adjusted in real time.
Thus, can be according to electrical equipment by being updated to the relation between history run state and historical failure situation
Running situation adjusts corresponding relation, is advantageous to be lifted the accuracy and reliability of failure predication.
In an optional example, the relation between the history run state and the historical failure situation is carried out more
Newly, can include:When the connection weight that this method can also include pair network structure related to the relation is adjusted,
Collect new service condition of the electrical equipment of same type in different user.
Alternatively, the relation between the history run state and the historical failure situation is updated, can be with
Including:Using new service condition as new sample data, the connection weight of the network structure is adjusted.
Alternatively, by new service condition and history service condition together as new sample data, to can be used for adjusting
After the regulation coefficient of the whole connection weight is adjusted, network structure described in re -training.
Such as:, can be using the new data collected in real time as new sample data, repeat step 3, weight when practicing
Newly network weight is adjusted, the accuracy rate of prediction can be improved.
Such as:Can also be using the data and historical data newly collected as sample data, the parameter η in adjustment type 2, again
Training network.
Thus, by being adjusted to the connection weight of network structure, the related tune of the connection weight for network structure
Integral coefficient is adjusted, it is possible to achieve the renewal of corresponding relation, and the mode updated is easy, reliable, the result of renewal precisely, peace
Quan Xing.
In an optional embodiment, it can also include:To the history run state, the historical failure situation,
At least one of the actual motion state, described physical fault situation are shown.
Thus, by the display to corresponding state, situation etc., user can be made to check corresponding data at any time, intuitive is strong,
Convenience is good.
Through substantial amounts of verification experimental verification, using the technical scheme of the present embodiment, by the generation of look-ahead air-conditioning failure, it is
Protection and maintenance is carried out to air-conditioning in time foundation is provided, the use of the infringement and the reduction of more lower bound degree of air-conditioner set to user can be lowered
Influence.
According to an embodiment of the invention, the failure for additionally providing a kind of electrical equipment of the failure prediction method corresponding to electrical equipment is pre-
Survey device.The structural representation of one embodiment of device of the invention shown in Figure 8.The fault prediction device of the electrical equipment can be with
Including:Control unit 102 and communication unit 104.
In an optional example, control unit 102, it can be used for the history run state and historical failure for establishing electrical equipment
Relation between situation.The concrete function of the control unit 102 and processing are referring to step S110.
Alternatively, the relation that described control unit 102 is established between the history run state of electrical equipment and historical failure situation,
It can specifically include:Collect history service condition of the electrical equipment of same type in different user.Wherein, the history uses feelings
Condition, it can include:History run state and historical failure situation during electric operation.The specific work(of the control unit 102
Can and it handle referring further to step S210.
Such as:By predicting the system and device of air-conditioning failure, service condition of the same class air-conditioning in different user is collected,
Including air-conditioning the various state parameters on running (containing operational factor, malfunction etc.).
More alternatively, in the history service condition, the history run state and the historical failure situation, press
Time sequencing arranges.
Such as:Air-conditioning in the previous year in service data collect, and arranged in chronological order.According to time sequence
Arrange obtained malfunction.
Thus, by being sequentially arranged history run state and historical failure situation, be advantageous to lifting and closed to corresponding
It is the accuracy determined, also helps the reliability of lifting failure predication.
More alternatively, described control unit 102 collects history service condition of the electrical equipment of same type in different user,
It can specifically include:By wireless transmitter, history of the electrical equipment of same type in preset duration in different user is received
Service condition.
Such as:The service data being collected into, the running state of air conditioner of a period of time is referred to, and be more than some moment
The state of point.
Thus, by collecting the history service condition in a period of time, be advantageous to obtain mass data and then lifted to going through
The accuracy and reliability that history running status and historical failure situation determine.
In an optional specific example, described control unit 102 establishes the history run state and historical failure of electrical equipment
Relation between situation, it can also specifically include:By big data analysis and digging technology, the history service condition is carried out
Analysis, obtains the data pair using the history run state as input parameter and using the historical failure situation as output parameter,
As sample data.The concrete function of the control unit 102 and processing are referring further to step S220.
Wherein, the input parameter, can include:Single history run state;And/or as setting rule described in
Individual features are extracted in history run state, and the one-dimension array or bidimensional above array being made up of the feature.
Such as:Input parameter can be not only single parameter, also include the input for extracting feature composition according to certain rules
Parameter one or more dimensions array.
Thus, by the input parameter of diversified forms, be advantageous to be lifted flexibility and the convenience of input mode.
More alternatively, described control unit 102 is analyzed the history service condition, can specifically be included:According to
At least one fault type of required prediction, learnt using SVMs, machine learning, at least one of data analysis side
Formula, the history service condition is analyzed, obtains analysis result.The concrete function of the control unit 102 and processing are also joined
See step S310.
In a more optional specific example, described control unit 102 is analyzed the history service condition, specifically
It can also include:With reference to default expertise, will have an impact in the analysis result to the fault type, and/or with institute
The related history run state of fault type is stated as input parameter, and by the analysis result with the input parameter phase
The historical failure situation answered is as output parameter.The concrete function of the control unit 102 and processing are referring further to step S320.
Such as:Some failures for needing to predict are determined in advance, by the analysis to data and combine expertise knowledge,
Choose respectively on that may have influence or relevant state parameter on these failures as input parameter, corresponding event
Barrier state is as output parameter.
Thus, by being analyzed by required fault type history service condition, and expertise, Ke Yiti are combined
Rise the accuracy determined to relation between history run state and historical failure situation and reliability.
In an optional specific example, described control unit 102 establishes the history run state and historical failure of electrical equipment
Relation between situation, it can also specifically include:Using neural network algorithm, the sample data is trained and tested,
To obtain the relation between the required history run state and the historical failure situation.The control unit 102 it is specific
Function and processing are referring further to step S230.
Such as:It is substantial amounts of by the air-conditioner set to collection using related algorithmic techniques such as big data analysis and excavations
History run state and failure situation are analyzed and excavated, using such as SVMs or other machines learning method or
Data analysing method, the non-linear relation between running state of air conditioner parameter and failure or rule are excavated and learnt, so as to
Algorithm foundation is provided for real-time failure predication.
Such as:For existing input and output set, such as SVMs or other machines learning method can be passed through
Or the relation of the analysis such as data analysing method input and outlet chamber, it is assumed that y is output, and (x can be one-dimensional or more to x for input
The array of group), existence function relation between y and x:Y=f (x).
Wherein, many some known variables in function f be present, and such as SVMs or other machines learn or number
According to analysis scheduling algorithm, this corresponding relation can be gone out according to existing input and output approximate simulation, that is, solve some unknown changes in f
The value of amount, obtain definite functional expression f.
Thus, by collecting the history service condition of electrical equipment, analyzed and excavated, obtain history in history service condition
Corresponding relation between running status and historical failure situation, be advantageous to be lifted the accuracy and reliability of failure predication.
More alternatively, described control unit 102 is trained and tested to the sample data, can specifically include:Will
The sample data is divided into training sample data and test sample data.The concrete function of the control unit 102 and processing are also
Referring to step S410.
Such as:Using the part in the sample data as training sample data, another part is as test sample number
According to.
Such as:By obtained input, output parameter pair, a part is used as training book sample data, and a part is used as test specimens
Notebook data.
In a more optional specific example, described control unit 102 is trained and tested to the sample data, tool
Body can also include:The network structure of neural network algorithm is chosen, initializes the connection weight of the network structure.The control list
The concrete function of member 102 and processing are referring further to step S420.
Wherein, the network structure, can include:In input node number, output node number, the network number of plies at least it
One.
Thus, by the network structure of diversified forms, training for promotion and the flexibility tested and versatility are advantageous to.
Such as:Choose network:According to the input of selection, output parameter, select LSTM network structures (such as:The network knot
Structure, it can include:Input and output nodal point number, network number of plies etc.), initialize network connection weight wk。
In a more optional specific example, described control unit 102 is trained and tested to the sample data, tool
Body can also include:By the network structure and the connection weight, the training sample data are trained.The control
The concrete function of unit 102 and processing are referring further to step S430.
Further, described control unit 102 is trained to the training sample data, can specifically be included:By institute
The input parameter for stating training sample data is input in the network structure, obtains theoretical output parameter.The control unit 102
Concrete function and processing are referring further to step S510.
In a further optional specific example, described control unit 102 is trained to the training sample data,
It can also specifically include:The training obtained between the theoretical output parameter and the output parameter of the training sample data misses
Difference, and determine the training error whether in the range of target training error.The concrete function of the control unit 102 and processing are also
Referring to step S520.
In a further optional specific example, described control unit 102 is trained to the training sample data,
It can also specifically include:When the training error is in the range of the target training error, deconditioning.The control unit
102 concrete function and processing are referring further to step S530.
In a further optional specific example, described control unit 102 is trained to the training sample data,
It can also specifically include:When the training error is outside the target training error scope, to the connection weight (such as:
wk) be adjusted.The concrete function of the control unit 102 and processing are referring further to step S540.
Such as:Training network:The input quantity of training sample data is input in network, obtains theoretical output valve a (x),
Corresponding real output value y (x) is compared to obtain error amount e (x)=‖ y (x)-a (x) ‖ with sample data.
If e (x) meets within desired value, i.e., | e (x) |<During ∈, deconditioning;Otherwise, according to Feedback error etc.
Method is to wkIt is adjusted.
Thus, by being trained to training sample data, required history run state and historical failure feelings can be obtained
Preliminary relationship between condition, reliability is high, and accuracy is good.
Wherein, described control unit 102 is adjusted to the connection weight, can specifically be included:According to the company
Connect the weights regulation coefficient related to the time series of the training sample (such as:Parameter η), pass through training error back transfer
Method is adjusted to the connection weight.
Such as:According to the methods of Feedback error to wkIt is adjusted:
C (w, b) is the error energy function (by taking standard variance function as an example) of training set, and n is the total quantity of training sample,
Summation is carried out on total training sample x:
Update weights:
Parameter η value can be preset in formula 2, can also be adjusted according to time series (for example time-sequencing is rearward
Training data, η values are larger, then role is larger).
Thus, by being adjusted to connection weight, reliability of the lifting to training network structure is advantageous to, and then can be with
The accuracy of training for promotion result.
In a more optional specific example, described control unit 102 is trained and tested to the sample data, tool
Body can also include:The test sample data are tested by the network structure and connection weight completed by the training.
The concrete function of the control unit 102 and processing are referring further to step S440.
Thus, by being trained to training sample data, then based on training result test sample data are tested,
It can obtain the corresponding relation of more accurate, relatively reliable history run state and historical failure situation, processing procedure is reliable,
Safety, processing mode are easy, accurate.
Further, described control unit 102 is tested the test sample data, can specifically be included:By institute
The input parameter for stating test sample data is inputted in the network structure, obtains theoretical output parameter.The tool of the control unit 102
Body function and processing are referring further to step S610.
In a further optional specific example, described control unit 102 is tested the test sample data,
It can also specifically include:The test obtained between the theoretical output parameter and the output parameter of the test sample data misses
Difference, and determine the test error whether in target detection error range.The concrete function of the control unit 102 and processing are also
Referring to step S620.
In a further optional specific example, described control unit 102 is tested the test sample data,
It can also specifically include:When the test error is in the target detection error range, stop test.The control unit
102 concrete function and processing are referring further to step S630.
In a further optional specific example, described control unit 102 is tested the test sample data,
It can also specifically include:When the test error is outside the target detection error range, network structure described in re -training;
And/or expand the capture range of history service condition of the electrical equipment of same type in different user, with the electricity to same type
History service condition of the device in different user is collected again.The concrete function of the control unit 102 and processing referring further to
Step S640.
Such as:The network that test sample data input has been trained, if the error energy function C of test set is unsatisfactory for
, it is necessary to which the step of repeating foregoing training network, if necessary, collects more data or the other input of selection again during preset value
Parameter.
Thus, by being tested on the basis of training result, to be verified to training result, lifting pair is advantageous to
The accuracy and reliability that relation between history run state and historical failure situation determines
In an optional example, communication unit 104, it can be used for the actual fortune for obtaining the electrical equipment for having established the relation
Row state.The concrete function of the communication unit 104 and processing are referring to step S120.
In an optional example, described control unit 102, can be also used for by the relation with the actual motion
The state historical failure situation corresponding to the history run state accordingly, as corresponding with the actual motion state
Physical fault situation, to obtain the failure predication result to electrical equipment.The concrete function of the control unit 102 and processing are referring further to step
Rapid S130.
Thus, by establishing the relation between history run state and historical failure situation, according to the relation according to reality
The failure that future may occur running status is predicted, with the generation of look-ahead air-conditioning failure, to enter in time to air-conditioning
Row protection and maintenance provides foundation, and can lower the use of the infringement and the reduction of more lower bound degree of air-conditioner set on user influences.
Wherein, the failure predication result, can include:Physical fault feelings from current time after a period of time
Condition.
Thus, by the prediction to following a period of time internal fault situation, electric operation situation can be grasped in advance, and then
Safeguarded in time when needing to safeguard, on the one hand can ensure electric operation safety, reliability height;On the other hand user can be ensured
Easy to use, hommization is good.
In an optional embodiment, it can also include:The process of fault pre-alarming.
In an optional example, described control unit 102, it can be also used for according to the failure predication result, it is determined that
Whether the fault degree of the physical fault situation reaches the early warning degree of setting.The concrete function of the control unit 102 and place
Reason is referring further to step S710.
Alternatively, described control unit 102, can be also used for when the fault degree of the physical fault situation reaches setting
Early warning degree when, initiate prompting, to prompt user to safeguard or intervene in time, so as to ensure the security of electric operation and user
The convenience used.The concrete function of the control unit 102 and processing are referring further to step S720.
Such as:Network has been trained, that is, has established y=f (x).Intelligent control center utilizes the network trained,
The data of the air-conditioning the past period arrived by the use of real-time collecting then obtain the failure shape of a period of time from now on as inputting
State, it will occur if prediction result is failure, information is delivered to terminal air-conditioning step by step, sending warning by air-conditioning reminds use
Report for repairment or ask for help in family.
Such as:Before air-conditioning substantially breaks down with regard to can prompt user or after sale personnel's air-conditioner set may occur therefore
Hinder and point out fault type.
Thus, by being prompted when failure needs to safeguard, to remind user to carry out Inspection and maintenance in time, and then lifted
The convenience that the security of electric operation and user use.
In an optional embodiment, described control unit 102, it can be also used for the history run state and institute
The relation stated between historical failure situation is updated.
Such as:The network algorithm structure of establishment, it is not unalterable, but can be adjusted in real time.
Thus, can be according to electrical equipment by being updated to the relation between history run state and historical failure situation
Running situation adjusts corresponding relation, is advantageous to be lifted the accuracy and reliability of failure predication.
In an optional example, described control unit 102 is to the history run state and the historical failure situation
Between relation be updated, can specifically include:When the device can also can be also used for pair including described control unit 102
When the connection weight of the network structure related to the relation is adjusted, the electrical equipment of same type is collected in different user
New service condition.
Alternatively, described control unit 102 is to the relation between the history run state and the historical failure situation
It is updated, can also specifically includes:Using new service condition as new sample data, to the connection weight of the network structure
Value is adjusted;And/or by new service condition and history service condition together as new sample data, to can be used for
Adjust the connection weight regulation coefficient be adjusted after, network structure described in re -training.
Such as:, can be using the new data collected in real time as new sample data, repeat step 3, weight when practicing
Newly network weight is adjusted, the accuracy rate of prediction can be improved.
Such as:Can also be using the data and historical data newly collected as sample data, the parameter η in adjustment type 2, again
Training network.
Thus, by being adjusted to the connection weight of network structure, the related tune of the connection weight for network structure
Integral coefficient is adjusted, it is possible to achieve the renewal of corresponding relation, and the mode updated is easy, reliable, the result of renewal precisely, peace
Quan Xing.
In an optional embodiment, described control unit 102, it can be also used for the history run state, institute
At least one of historical failure situation, the actual motion state, described physical fault situation is stated to be shown.
Thus, by the display to corresponding state, situation etc., user can be made to check corresponding data at any time, intuitive is strong,
Convenience is good.
The processing and function realized by the device of the present embodiment essentially correspond to earlier figures 1 to the method shown in Fig. 7
Embodiment, principle and example, therefore not detailed part in the description of the present embodiment may refer to mutually speaking on somebody's behalf in previous embodiment
It is bright, it will not be described here.
Through substantial amounts of verification experimental verification, using technical scheme, transported by the substantial amounts of history of the air-conditioner set of collection
Row state and failure situation are analyzed and excavated, and establish the running status of air-conditioning and the relation of failure situation, for pre- in advance
Survey failure generation, early to air-conditioning carry out Inspection and maintenance, then can lower air-conditioner set infringement and more lower bound degree reduction to
The use at family influences.
According to an embodiment of the invention, a kind of storage medium of the failure prediction method corresponding to electrical equipment is additionally provided.Should
Storage medium, it can include:A plurality of instruction is stored with the storage medium;The a plurality of instruction, for being loaded by processor
And perform the failure prediction method of above-described electrical equipment.
The processing and function realized by the storage medium of the present embodiment essentially correspond to earlier figures 1 to shown in Fig. 7
Embodiment, principle and the example of method, therefore not detailed part in the description of the present embodiment, may refer to the phase in previous embodiment
Speak on somebody's behalf bright, will not be described here.
Through substantial amounts of verification experimental verification, using technical scheme, by before air-conditioning substantially breaks down just
User can be prompted or personnel's air-conditioner set may break down and point out fault type after sale, can lower air-conditioner set infringement and
The use that more lower bound degree reduces on user influences.
According to an embodiment of the invention, a kind of electrical equipment of the failure prediction method corresponding to electrical equipment is additionally provided.The electrical equipment,
It can include:Processor, for performing a plurality of instruction;Memory, for storing a plurality of instruction;Wherein, a plurality of instruction, use
In by the memory storage, and loaded by the processor and perform the failure prediction method of above-described electrical equipment.Or
The electrical equipment, it can include:The fault prediction device of above-described electrical equipment.
Alternatively, the electrical equipment, can include:In air-conditioning, refrigerator, washing machine, smoke exhaust ventilator, water heater, air purifier
At least one of.
In an optional embodiment, the scheme of the air-conditioning failure predication, it can include:Analyzed and dug using big data
The related algorithmic technique such as pick, by the substantial amounts of history run state of air-conditioner set to collection and failure situation carry out analysis and
Excavate, using such as SVMs or other machines learning method or data analysing method, to running state of air conditioner parameter
Non-linear relation or rule between failure are excavated and learnt, so as to provide algorithm foundation for real-time failure predication.
In an optional example, failure predication can be carried out by a kind of system and device for predicting air-conditioning failure.Such as Fig. 9
Shown, the air-conditioning with wireless transmitter, through router, either mobile phone or other-end are connected with internet, are connected to
Cloud server, be eventually connected to intelligent control center (such as:The intelligent control center, can be user side or service provider side
Built-in system).The device being made up of the series module, can by the running situation of different terminals air-conditioning (including operational factor,
Malfunction etc.) pass intelligent control center back, relevant control can also be instructed (such as indicating fault) by intelligent control center
It is dealt on terminal air-conditioning.
In an optional example, the implementation process of the technology of the air-conditioning failure predication, it can include:
1st, data collection:
By predicting the system and device of air-conditioning failure, service condition of the same class air-conditioning in different user is collected, including
Various state parameters of the air-conditioning on running (containing operational factor, malfunction etc.).
Here service data refers to the running state of air conditioner of a period of time, and is more than the shape of some moment point
State.For example, air-conditioning in the previous year in service data collect, and arranged in chronological order.Obtained according to time series
The malfunction arrived.
2nd, sample data selects
Some failures for needing to predict are determined in advance, by the analysis to data and combine expertise knowledge, respectively
Choose on that may have influence or relevant state parameter as input parameter, corresponding failure shape to these failures
State is as output parameter.
Input parameter can be not only single parameter, also include the input parameter one for extracting feature composition according to certain rules
Dimension or Multidimensional numerical.
By obtained input, output parameter pair, a part is used as training book sample data, and a part is used as test sample number
According to.
3rd, prediction algorithm is chosen and debugged
For existing input and output set, such as SVMs or other machines learning method or number can be passed through
According to the analysis such as analysis method input and the relation of outlet chamber, it is assumed that y is output, and for input, (x can be one-dimensional or multigroup number to x
Group), existence function relation between y and x:Y=f (x).
Wherein, many some known variables in function f be present, and such as SVMs or other machines learn or number
According to analysis scheduling algorithm, this corresponding relation can be gone out according to existing input and output approximate simulation, that is, solve some unknown changes in f
The value of amount, obtain definite functional expression f.
Alternatively, specific algorithm implementation has many kinds, and the application can not be enumerated, is only illustrated below:
LSTM (Long Short-Term Memory, shot and long term memory network) neutral net in deep learning network,
Having to solution to sequential has powerful ability on the model of dependence, be suitable in status predication.Theory on LSTM networks
Pattern, training method etc., the corresponding description of existing books or paper can be utilized, is not described in detail herein.
3.1 choose network:According to the input of selection, output parameter, select LSTM network structures (such as:The network structure,
It can include:Input and output nodal point number, network number of plies etc.), initialize network connection weight wk。
3.2 training network:The input quantity of training sample data is input in network, obtains theoretical output valve a (x), and
Corresponding real output value y (x) is compared to obtain error amount e (x)=‖ y (x)-a (x) ‖ in sample data.
If e (x) meets within desired value, i.e., | e (x) |<During ∈, deconditioning;Otherwise, according to Feedback error etc.
Method is to wkIt is adjusted:
C (w, b) is the error energy function (by taking standard variance function as an example) of training set, and n is the total quantity of training sample,
Summation is carried out on total training sample x:
Update weights:
Parameter η value can be preset in formula 2, can also be adjusted according to time series (for example time-sequencing is rearward
Training data, η values are larger, then role is larger).
The method of network training has many kinds, and only citing is a kind of herein.
3.3 test network
The network that will have been trained in test sample data input 3.2, if the error energy function C of test set is unsatisfactory for
, it is necessary to repeat above-mentioned 3.1~3.2 step during preset value, if necessary, more data or the other input ginseng of selection are collected again
Number.
4th, failure predication practices the stage
In step 3, network is trained, that is, has established y=f (x).Intelligent control center utilizes what is trained
Network, the data of the air-conditioning the past period arrived by the use of real-time collecting then obtain the event of a period of time from now on as input
Barrier state, it will occur if prediction result is failure, and information will be delivered to terminal air-conditioning step by step, sending warning by air-conditioning will carry
Awake user reports for repairment or asked for help.
5th, algorithm on-line tuning and amendment stage
The network algorithm structure of the inner establishment of step 3, it is not unalterable, but can be adjusted in real time.
, can the new data that collected in real time is again right as new sample data, repeat step 3 when practicing
Network weight is adjusted, and can improve the accuracy rate of prediction.
Can also be using the data and historical data newly collected as sample data, the parameter η in adjustment type 2, re -training
Network.
It can be seen that the scheme of the air-conditioning failure predication, can solve or at least partly solve problems of the prior art,
I.e. before air-conditioning substantially breaks down with regard to can prompt user or after sale personnel's air-conditioner set may break down and point out therefore
Hinder type.
Such as:The failure of many types, refrigerant lacks failure, motor corrupted failure etc., is the mistake of a gradual change
Journey, it is that air conditioner operation parameters are gradually deviated from normal operating condition until unit warning is out of order, can not reruned.The air-conditioning failure
The scheme of prediction, what is solved is exactly the process that identify this gradual change, and failure is predicted.
The processing and function realized by the electrical equipment of the present embodiment essentially correspond to earlier figures 1 to the method shown in Fig. 7,
Or embodiment, principle and the example of the device shown in earlier figures 8, therefore not detailed part in the description of the present embodiment, it may refer to
Related description in previous embodiment, will not be described here.
Through substantial amounts of verification experimental verification, using technical scheme, pass through the running status to air-conditioning and failure situation
Relation safeguarded and updated, the accuracy rate of prediction can be improved.
To sum up, it will be readily appreciated by those skilled in the art that on the premise of not conflicting, above-mentioned each advantageous manner can be certainly
Combined, be superimposed by ground.
Embodiments of the invention are the foregoing is only, are not intended to limit the invention, for those skilled in the art
For member, the present invention can have various modifications and variations.Any modification within the spirit and principles of the invention, being made,
Equivalent substitution, improvement etc., should be included within scope of the presently claimed invention.
Claims (17)
- A kind of 1. failure prediction method of electrical equipment, it is characterised in that including:The relation established between the history run state of electrical equipment and historical failure situation;Obtain the actual motion state for the electrical equipment for having established the relation;By the historical failure feelings corresponding to the history run state corresponding with the actual motion state in the relation Condition, as physical fault situation corresponding with the actual motion state, to obtain the failure predication result to electrical equipment.
- 2. according to the method for claim 1, it is characterised in that wherein,The relation established between the history run state of electrical equipment and historical failure situation, including:Collect history service condition of the electrical equipment of same type in different user;Wherein, the history service condition, including: History run state and historical failure situation during electric operation;By big data analysis and digging technology, the history service condition is analyzed, obtained with the history run shape State is input parameter and the data pair using the historical failure situation as output parameter, as sample data;Using neural network algorithm, the sample data is trained and tested, to obtain the required history run shape Relation between state and the historical failure situation.
- 3. according to the method for claim 2, it is characterised in that wherein,History service condition of the electrical equipment of same type in different user is collected, including:By wireless transmitter, history service condition of the electrical equipment of same type in preset duration in different user is received;And/orThe history service condition is analyzed, including:According at least one fault type of required prediction, learnt using SVMs, machine learning, in data analysis extremely A kind of few mode, analyzes the history service condition, obtains analysis result;With reference to default expertise, will have an impact in the analysis result to the fault type, and/or with the failure classes The related history run state of type is as input parameter, and by history corresponding with the input parameter in the analysis result Failure situation is as output parameter;And/orThe sample data is trained and tested, including:The sample data is divided into training sample data and test sample data;The network structure of neural network algorithm is chosen, initializes the connection weight of the network structure;By the network structure and the connection weight, the training sample data are trained;The test sample data are tested by the network structure and connection weight completed by the training.
- 4. according to the method for claim 3, it is characterised in that wherein,The training sample data are trained, including:The input parameter of the training sample data is input in the network structure, obtains theoretical output parameter;The training error between the theoretical output parameter and the output parameter of the training sample data is obtained, and described in determination Whether training error is in the range of target training error;When the training error is in the range of the target training error, deconditioning;Or when the training error is outside the target training error scope, the connection weight is adjusted;And/orThe test sample data are tested, including:The input parameter of the test sample data is inputted in the network structure, obtains theoretical output parameter;The test error between the theoretical output parameter and the output parameter of the test sample data is obtained, and described in determination Whether test error is in target detection error range;When the test error is in the target detection error range, stop test;Or when the test error is outside the target detection error range, network structure described in re -training;And/or expand The capture range of history service condition of the electrical equipment of big same type in different user, with the electrical equipment to same type in difference History service condition in user is collected again.
- 5. according to the method for claim 4, it is characterised in that wherein,In the history service condition, the history run state and the historical failure situation, it is sequentially arranged;And/orThe input parameter, including:Single history run state;And/orIndividual features, and the one-dimension array being made up of the feature or two are extracted from the history run state by setting rule Tie up above array;And/orThe network structure, including:At least one of input node number, output node number, network number of plies;And/orThe connection weight is adjusted, including:It is reverse by training error according to the regulation coefficient related to the time series of the connection weight and the training sample TRANSFER METHOD is adjusted to the connection weight.
- 6. according to the method described in one of claim 1-5, it is characterised in that also include:According to the failure predication result, determine whether the fault degree of the physical fault situation reaches the early warning journey of setting Degree;When the fault degree of the physical fault situation reaches the early warning degree of setting, prompting is initiated;And/orRelation between the history run state and the historical failure situation is updated;And/orTo in the history run state, the historical failure situation, the actual motion state, the physical fault situation At least one shown.
- 7. according to the method for claim 6, it is characterised in that wherein,Relation between the history run state and the historical failure situation is updated, including:When also the connection weight including pair network structure related to the relation is adjusted this method, same type is collected New service condition of the electrical equipment in different user;Using new service condition as new sample data, the connection weight of the network structure is adjusted;And/orBy new service condition and history service condition together as new sample data, to for adjusting the connection weight After regulation coefficient is adjusted, network structure described in re -training;And/orThe failure predication result, including:Physical fault situation from current time after a period of time.
- A kind of 8. fault prediction device of electrical equipment, it is characterised in that including:Control unit, for establishing the relation between the history run state of electrical equipment and historical failure situation;Communication unit, for obtaining the actual motion state for the electrical equipment for having established the relation;Described control unit, it is additionally operable to the history run state pair corresponding with the actual motion state in the relation The historical failure situation answered, as physical fault situation corresponding with the actual motion state, to obtain to electrical equipment Failure predication result.
- 9. device according to claim 8, it is characterised in that wherein,The relation that described control unit is established between the history run state of electrical equipment and historical failure situation, is specifically included:Collect history service condition of the electrical equipment of same type in different user;Wherein, the history service condition, including: History run state and historical failure situation during electric operation;By big data analysis and digging technology, the history service condition is analyzed, obtained with the history run shape State is input parameter and the data pair using the historical failure situation as output parameter, as sample data;Using neural network algorithm, the sample data is trained and tested, to obtain the required history run shape Relation between state and the historical failure situation.
- 10. device according to claim 9, it is characterised in that wherein,Described control unit collects history service condition of the electrical equipment of same type in different user, specifically includes:By wireless transmitter, history service condition of the electrical equipment of same type in preset duration in different user is received;And/orDescribed control unit is analyzed the history service condition, is specifically included:According at least one fault type of required prediction, learnt using SVMs, machine learning, in data analysis extremely A kind of few mode, analyzes the history service condition, obtains analysis result;With reference to default expertise, will have an impact in the analysis result to the fault type, and/or with the failure classes The related history run state of type is as input parameter, and by history corresponding with the input parameter in the analysis result Failure situation is as output parameter;And/orDescribed control unit is trained and tested to the sample data, specifically includes:The sample data is divided into training sample data and test sample data;The network structure of neural network algorithm is chosen, initializes the connection weight of the network structure;By the network structure and the connection weight, the training sample data are trained;The test sample data are tested by the network structure and connection weight completed by the training.
- 11. device according to claim 10, it is characterised in that wherein,Described control unit is trained to the training sample data, is specifically included:The input parameter of the training sample data is input in the network structure, obtains theoretical output parameter;The training error between the theoretical output parameter and the output parameter of the training sample data is obtained, and described in determination Whether training error is in the range of target training error;When the training error is in the range of the target training error, deconditioning;Or when the training error is outside the target training error scope, the connection weight is adjusted;And/orDescribed control unit is tested the test sample data, is specifically included:The input parameter of the test sample data is inputted in the network structure, obtains theoretical output parameter;The test error between the theoretical output parameter and the output parameter of the test sample data is obtained, and described in determination Whether test error is in target detection error range;When the test error is in the target detection error range, stop test;Or when the test error is outside the target detection error range, network structure described in re -training;And/or expand The capture range of history service condition of the electrical equipment of big same type in different user, with the electrical equipment to same type in difference History service condition in user is collected again.
- 12. device according to claim 11, it is characterised in that wherein,In the history service condition, the history run state and the historical failure situation, it is sequentially arranged;And/orThe input parameter, including:Single history run state;And/orIndividual features, and the one-dimension array being made up of the feature or two are extracted from the history run state by setting rule Tie up above array;And/orThe network structure, including:At least one of input node number, output node number, network number of plies;And/orDescribed control unit is adjusted to the connection weight, is specifically included:It is reverse by training error according to the regulation coefficient related to the time series of the connection weight and the training sample TRANSFER METHOD is adjusted to the connection weight.
- 13. according to the device described in one of claim 8-12, it is characterised in that also include:Described control unit, the fault degree for being additionally operable to, according to the failure predication result, determine the physical fault situation are The no early warning degree for reaching setting;When the fault degree of the physical fault situation reaches the early warning degree of setting, prompting is initiated;And/orDescribed control unit, it is additionally operable to carry out more the relation between the history run state and the historical failure situation Newly;And/orDescribed control unit, be additionally operable to the history run state, the historical failure situation, the actual motion state, At least one of described physical fault situation is shown.
- 14. device according to claim 13, it is characterised in that wherein,Described control unit is updated to the relation between the history run state and the historical failure situation, specific bag Include:The connection weight for being also additionally operable to pair network structure related to the relation including described control unit when the device is carried out During adjustment, new service condition of the electrical equipment of same type in different user is collected;Using new service condition as new sample data, the connection weight of the network structure is adjusted;And/orBy new service condition and history service condition together as new sample data, to for adjusting the connection weight After regulation coefficient is adjusted, network structure described in re -training;And/orThe failure predication result, including:Physical fault situation from current time after a period of time.
- 15. a kind of storage medium, it is characterised in that a plurality of instruction is stored with the storage medium;The a plurality of instruction, is used for Loaded by processor and perform the failure prediction method of the electrical equipment as described in claim 1-7 is any.
- A kind of 16. electrical equipment, it is characterised in that including:Processor, for performing a plurality of instruction;Memory, for storing a plurality of instruction;Wherein, a plurality of instruction, for loading and performing such as claim by the memory storage, and by the processor The failure prediction method of any described electrical equipment of 1-7;OrThe fault prediction device of electrical equipment as described in claim 8-14 is any.
- 17. electrical equipment according to claim 16, it is characterised in that the electrical equipment, including:Air-conditioning, refrigerator, washing machine, take out At least one of lampblack absorber, water heater, air purifier.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710824151.2A CN107679649A (en) | 2017-09-13 | 2017-09-13 | A kind of failure prediction method of electrical equipment, device, storage medium and electrical equipment |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN201710824151.2A CN107679649A (en) | 2017-09-13 | 2017-09-13 | A kind of failure prediction method of electrical equipment, device, storage medium and electrical equipment |
Publications (1)
Publication Number | Publication Date |
---|---|
CN107679649A true CN107679649A (en) | 2018-02-09 |
Family
ID=61135752
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN201710824151.2A Pending CN107679649A (en) | 2017-09-13 | 2017-09-13 | A kind of failure prediction method of electrical equipment, device, storage medium and electrical equipment |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN107679649A (en) |
Cited By (35)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108564280A (en) * | 2018-04-13 | 2018-09-21 | 上海财经大学 | Worksheet processing classification maintenance unit based on history maintenance record |
CN109034413A (en) * | 2018-07-11 | 2018-12-18 | 广东人励智能工程有限公司 | Intelligence manufacture equipment fault prediction technique and system based on neural network model |
CN109063854A (en) * | 2018-08-23 | 2018-12-21 | 河南中裕广恒科技股份有限公司 | Intelligent O&M cloud platform system and its control method |
CN109739213A (en) * | 2019-01-07 | 2019-05-10 | 东莞百宏实业有限公司 | A kind of failure prediction system and prediction technique |
CN109946544A (en) * | 2019-03-29 | 2019-06-28 | 广东美的制冷设备有限公司 | Household electrical appliances fault detection method and device |
CN110207321A (en) * | 2019-06-10 | 2019-09-06 | 珠海格力电器股份有限公司 | Fault handling method, equipment and the air-conditioning system of self-regulation |
CN110297179A (en) * | 2018-05-11 | 2019-10-01 | 宫文峰 | Diesel-driven generator failure predication and monitoring system device based on integrated deep learning |
CN110297178A (en) * | 2018-05-11 | 2019-10-01 | 宫文峰 | Diesel generating set fault diagnosis and detection device and method based on deep learning |
CN110425694A (en) * | 2019-08-09 | 2019-11-08 | 长江慧控科技(武汉)有限公司 | High-speed rail intelligence station Heating,Ventilating and Air Conditioning efficiency based on PHM controls management method |
CN110500709A (en) * | 2019-07-29 | 2019-11-26 | 七彩安科智慧科技有限公司 | The online method of discrimination of conditioners running conditions |
CN110515935A (en) * | 2019-08-29 | 2019-11-29 | 新誉轨道交通科技有限公司 | A kind of dirty stifled failure prediction method of train air-conditioning strainer, device, equipment and system |
CN110738827A (en) * | 2018-07-20 | 2020-01-31 | 珠海格力电器股份有限公司 | Abnormity early warning method, system, device and storage medium of electric appliance |
CN110836403A (en) * | 2018-08-17 | 2020-02-25 | 珠海格力电器股份有限公司 | Fault detection method and device of range hood |
CN110837229A (en) * | 2018-08-17 | 2020-02-25 | 珠海格力电器股份有限公司 | Control method and device for household appliance |
CN110895721A (en) * | 2018-09-12 | 2020-03-20 | 珠海格力电器股份有限公司 | Method and device for predicting electric appliance function |
CN111027591A (en) * | 2019-11-13 | 2020-04-17 | 西安交通大学 | Node fault prediction method for large-scale cluster system |
CN111105063A (en) * | 2018-10-26 | 2020-05-05 | 北京国双科技有限公司 | Fault prediction method, fault prediction device, model construction method, fault prediction device, processor and readable storage medium |
CN111123087A (en) * | 2019-12-24 | 2020-05-08 | 深圳供电局有限公司 | Power distribution network switching equipment fault detection method and device and computer equipment |
CN111174370A (en) * | 2018-11-09 | 2020-05-19 | 珠海格力电器股份有限公司 | Fault detection method and device, storage medium and electronic device |
CN111289817A (en) * | 2020-02-14 | 2020-06-16 | 珠海格力电器股份有限公司 | Method, device and system for monitoring faults of electric appliance and storage medium |
CN111309562A (en) * | 2020-02-27 | 2020-06-19 | 上海上大鼎正软件股份有限公司 | Server failure prediction method, device, equipment and storage medium |
CN111476400A (en) * | 2020-03-11 | 2020-07-31 | 珠海格力电器股份有限公司 | Circuit fault prediction method, device, equipment and computer readable medium |
CN111578444A (en) * | 2019-02-19 | 2020-08-25 | 珠海格力电器股份有限公司 | Air conditioner fault prediction method and device, storage medium and air conditioner |
CN111739622A (en) * | 2020-07-22 | 2020-10-02 | 北京梦天门科技股份有限公司 | Medical waste on-line monitoring and predicting method and system |
CN112445193A (en) * | 2019-08-14 | 2021-03-05 | 珠海格力电器股份有限公司 | Method, device and equipment for predicting air conditioner fault and storage medium |
EP3789838A1 (en) * | 2019-09-09 | 2021-03-10 | Alisea S.r.l. | Systems and methods for artificial intelligence-based maintenance of an air conditioning system |
CN112803581A (en) * | 2021-03-07 | 2021-05-14 | 广东电网有限责任公司广州供电局 | Intelligent power supply control method, system and device of low-voltage contact cabinet |
CN112966379A (en) * | 2021-03-07 | 2021-06-15 | 广东电网有限责任公司广州供电局 | Anti-misoperation method and device based on locking |
CN113283646A (en) * | 2021-05-14 | 2021-08-20 | 青岛海尔科技有限公司 | Method and device for controlling equipment operation, storage medium and electronic device |
CN113819575A (en) * | 2021-08-18 | 2021-12-21 | 青岛海尔空调器有限总公司 | Control method and device for air conditioner and server |
WO2023020078A1 (en) * | 2021-08-18 | 2023-02-23 | 青岛海尔空调器有限总公司 | Control method and device for air conditioner, and server |
WO2023094064A1 (en) * | 2021-11-25 | 2023-06-01 | Viessmann Climate Solutions Se | Method for predicting faults in an hvac system |
CN116611006A (en) * | 2023-05-22 | 2023-08-18 | 广州吉谷电器有限公司 | Fault identification method and device of electric kettle based on user feedback |
CN117390468A (en) * | 2023-10-19 | 2024-01-12 | 应急管理部大数据中心 | Dust removal equipment state monitoring and early warning method and system based on data mining |
CN117390468B (en) * | 2023-10-19 | 2024-06-04 | 应急管理部大数据中心 | Dust removal equipment state monitoring and early warning method and system based on data mining |
Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103138980A (en) * | 2011-11-30 | 2013-06-05 | 海尔集团公司 | Equipment failure feed-back system and method |
KR101398413B1 (en) * | 2013-08-19 | 2014-05-27 | 한국전기안전공사 | Method for learning and calculating the number of received cases of electrical fault using weather data and neural networks |
CN104951836A (en) * | 2014-03-25 | 2015-09-30 | 上海市玻森数据科技有限公司 | Posting predication system based on nerual network technique |
CN105095963A (en) * | 2015-08-17 | 2015-11-25 | 中国空气动力研究与发展中心高速空气动力研究所 | Method for accurately diagnosing and predicting fault of wind tunnel equipment |
CN106096170A (en) * | 2016-06-21 | 2016-11-09 | 山西大学 | Wind turbines multivariate failure prediction method based on data-driven |
-
2017
- 2017-09-13 CN CN201710824151.2A patent/CN107679649A/en active Pending
Patent Citations (5)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN103138980A (en) * | 2011-11-30 | 2013-06-05 | 海尔集团公司 | Equipment failure feed-back system and method |
KR101398413B1 (en) * | 2013-08-19 | 2014-05-27 | 한국전기안전공사 | Method for learning and calculating the number of received cases of electrical fault using weather data and neural networks |
CN104951836A (en) * | 2014-03-25 | 2015-09-30 | 上海市玻森数据科技有限公司 | Posting predication system based on nerual network technique |
CN105095963A (en) * | 2015-08-17 | 2015-11-25 | 中国空气动力研究与发展中心高速空气动力研究所 | Method for accurately diagnosing and predicting fault of wind tunnel equipment |
CN106096170A (en) * | 2016-06-21 | 2016-11-09 | 山西大学 | Wind turbines multivariate failure prediction method based on data-driven |
Cited By (42)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN108564280A (en) * | 2018-04-13 | 2018-09-21 | 上海财经大学 | Worksheet processing classification maintenance unit based on history maintenance record |
CN110297179A (en) * | 2018-05-11 | 2019-10-01 | 宫文峰 | Diesel-driven generator failure predication and monitoring system device based on integrated deep learning |
CN110297178A (en) * | 2018-05-11 | 2019-10-01 | 宫文峰 | Diesel generating set fault diagnosis and detection device and method based on deep learning |
CN109034413A (en) * | 2018-07-11 | 2018-12-18 | 广东人励智能工程有限公司 | Intelligence manufacture equipment fault prediction technique and system based on neural network model |
CN110738827A (en) * | 2018-07-20 | 2020-01-31 | 珠海格力电器股份有限公司 | Abnormity early warning method, system, device and storage medium of electric appliance |
CN110836403A (en) * | 2018-08-17 | 2020-02-25 | 珠海格力电器股份有限公司 | Fault detection method and device of range hood |
CN110837229A (en) * | 2018-08-17 | 2020-02-25 | 珠海格力电器股份有限公司 | Control method and device for household appliance |
CN109063854A (en) * | 2018-08-23 | 2018-12-21 | 河南中裕广恒科技股份有限公司 | Intelligent O&M cloud platform system and its control method |
CN110895721A (en) * | 2018-09-12 | 2020-03-20 | 珠海格力电器股份有限公司 | Method and device for predicting electric appliance function |
CN111105063A (en) * | 2018-10-26 | 2020-05-05 | 北京国双科技有限公司 | Fault prediction method, fault prediction device, model construction method, fault prediction device, processor and readable storage medium |
CN111174370A (en) * | 2018-11-09 | 2020-05-19 | 珠海格力电器股份有限公司 | Fault detection method and device, storage medium and electronic device |
CN109739213A (en) * | 2019-01-07 | 2019-05-10 | 东莞百宏实业有限公司 | A kind of failure prediction system and prediction technique |
CN111578444A (en) * | 2019-02-19 | 2020-08-25 | 珠海格力电器股份有限公司 | Air conditioner fault prediction method and device, storage medium and air conditioner |
CN109946544A (en) * | 2019-03-29 | 2019-06-28 | 广东美的制冷设备有限公司 | Household electrical appliances fault detection method and device |
CN110207321A (en) * | 2019-06-10 | 2019-09-06 | 珠海格力电器股份有限公司 | Fault handling method, equipment and the air-conditioning system of self-regulation |
CN110207321B (en) * | 2019-06-10 | 2020-10-02 | 珠海格力电器股份有限公司 | Self-adjusting fault processing method and equipment and air conditioning system |
CN110500709A (en) * | 2019-07-29 | 2019-11-26 | 七彩安科智慧科技有限公司 | The online method of discrimination of conditioners running conditions |
CN110425694A (en) * | 2019-08-09 | 2019-11-08 | 长江慧控科技(武汉)有限公司 | High-speed rail intelligence station Heating,Ventilating and Air Conditioning efficiency based on PHM controls management method |
CN112445193A (en) * | 2019-08-14 | 2021-03-05 | 珠海格力电器股份有限公司 | Method, device and equipment for predicting air conditioner fault and storage medium |
CN110515935A (en) * | 2019-08-29 | 2019-11-29 | 新誉轨道交通科技有限公司 | A kind of dirty stifled failure prediction method of train air-conditioning strainer, device, equipment and system |
EP3789838A1 (en) * | 2019-09-09 | 2021-03-10 | Alisea S.r.l. | Systems and methods for artificial intelligence-based maintenance of an air conditioning system |
US11156997B2 (en) | 2019-09-09 | 2021-10-26 | Alisea S.R.L. | Systems and methods for artificial intelligence-based maintenance of an air conditioning system |
CN111027591A (en) * | 2019-11-13 | 2020-04-17 | 西安交通大学 | Node fault prediction method for large-scale cluster system |
CN111027591B (en) * | 2019-11-13 | 2022-07-12 | 西安交通大学 | Node fault prediction method for large-scale cluster system |
CN111123087A (en) * | 2019-12-24 | 2020-05-08 | 深圳供电局有限公司 | Power distribution network switching equipment fault detection method and device and computer equipment |
CN111289817A (en) * | 2020-02-14 | 2020-06-16 | 珠海格力电器股份有限公司 | Method, device and system for monitoring faults of electric appliance and storage medium |
CN111309562A (en) * | 2020-02-27 | 2020-06-19 | 上海上大鼎正软件股份有限公司 | Server failure prediction method, device, equipment and storage medium |
CN111309562B (en) * | 2020-02-27 | 2023-07-28 | 上海上大鼎正软件股份有限公司 | Method, device, equipment and storage medium for predicting server faults |
CN111476400A (en) * | 2020-03-11 | 2020-07-31 | 珠海格力电器股份有限公司 | Circuit fault prediction method, device, equipment and computer readable medium |
CN111739622B (en) * | 2020-07-22 | 2021-01-26 | 北京梦天门科技股份有限公司 | Medical waste on-line monitoring and predicting method and system |
CN111739622A (en) * | 2020-07-22 | 2020-10-02 | 北京梦天门科技股份有限公司 | Medical waste on-line monitoring and predicting method and system |
CN112803581A (en) * | 2021-03-07 | 2021-05-14 | 广东电网有限责任公司广州供电局 | Intelligent power supply control method, system and device of low-voltage contact cabinet |
CN112966379A (en) * | 2021-03-07 | 2021-06-15 | 广东电网有限责任公司广州供电局 | Anti-misoperation method and device based on locking |
CN113283646A (en) * | 2021-05-14 | 2021-08-20 | 青岛海尔科技有限公司 | Method and device for controlling equipment operation, storage medium and electronic device |
WO2023020081A1 (en) * | 2021-08-18 | 2023-02-23 | 青岛海尔空调器有限总公司 | Method and apparatus for controlling air conditioner, and server |
WO2023020078A1 (en) * | 2021-08-18 | 2023-02-23 | 青岛海尔空调器有限总公司 | Control method and device for air conditioner, and server |
CN113819575A (en) * | 2021-08-18 | 2021-12-21 | 青岛海尔空调器有限总公司 | Control method and device for air conditioner and server |
WO2023094064A1 (en) * | 2021-11-25 | 2023-06-01 | Viessmann Climate Solutions Se | Method for predicting faults in an hvac system |
CN116611006A (en) * | 2023-05-22 | 2023-08-18 | 广州吉谷电器有限公司 | Fault identification method and device of electric kettle based on user feedback |
CN116611006B (en) * | 2023-05-22 | 2024-02-20 | 广州吉谷电器有限公司 | Fault identification method and device of electric kettle based on user feedback |
CN117390468A (en) * | 2023-10-19 | 2024-01-12 | 应急管理部大数据中心 | Dust removal equipment state monitoring and early warning method and system based on data mining |
CN117390468B (en) * | 2023-10-19 | 2024-06-04 | 应急管理部大数据中心 | Dust removal equipment state monitoring and early warning method and system based on data mining |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107679649A (en) | A kind of failure prediction method of electrical equipment, device, storage medium and electrical equipment | |
CN104866632B (en) | Fault data acquisition methods, device and the terminal of home appliance | |
CN110929918B (en) | 10kV feeder fault prediction method based on CNN and LightGBM | |
TWI385492B (en) | A system for maintaining and analyzing manufacturing equipment and method therefor | |
CN106527339B (en) | A kind of highly reliable preparation equipment fault diagnosis system and method based on industrial cloud | |
CN104732276B (en) | One kind metering production facility on-line fault diagnosis method | |
CN105467971B (en) | A kind of second power equipment monitoring system and method | |
CN105425768B (en) | A kind of second power equipment monitoring device and method | |
CN111412579B (en) | Air conditioning unit fault type diagnosis method and system based on big data | |
CN108664010A (en) | Generating set fault data prediction technique, device and computer equipment | |
CN107860099A (en) | Frosting detection method, device, storage medium and the equipment of a kind of air-conditioning | |
CN104254809A (en) | Method and system for diagnostic rules for heavy duty gas turbines | |
CN109501834A (en) | A kind of point machine failure prediction method and device | |
CN107770797A (en) | A kind of association analysis method and system of wireless network alarm management | |
CN108051709A (en) | Transformer state online evaluation analysis method based on artificial intelligence technology | |
KR101660102B1 (en) | Apparatus for water demand forecasting | |
CN109685389A (en) | Elevator faults work dispatching method, device, server, storage medium and system | |
CN107295537A (en) | A kind of method and system for wireless sensor network reliability of testing and assessing | |
CN116418117A (en) | Equipment detection system for intelligent power grid | |
CN111444075A (en) | Method for automatically discovering key influence indexes | |
CN109284886A (en) | Electrical Safety management method and device based on artificial intelligence | |
CN107305238A (en) | The method for diagnosing faults and device of a kind of main exhauster of sintering | |
CN112926756A (en) | Application method of AI (AI) maintenance knowledge base of central air-conditioning equipment | |
US20140052425A1 (en) | Method and apparatus for evaluating a model of an industrial plant process | |
CN114471170B (en) | Ceramic membrane automatic dosing cleaning system and method based on Internet of things |
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 | ||
RJ01 | Rejection of invention patent application after publication |
Application publication date: 20180209 |
|
RJ01 | Rejection of invention patent application after publication |