CN109100975A - A kind of parameter optimization method and system - Google Patents

A kind of parameter optimization method and system Download PDF

Info

Publication number
CN109100975A
CN109100975A CN201811024483.3A CN201811024483A CN109100975A CN 109100975 A CN109100975 A CN 109100975A CN 201811024483 A CN201811024483 A CN 201811024483A CN 109100975 A CN109100975 A CN 109100975A
Authority
CN
China
Prior art keywords
parameter
model
controllable parameter
controllable
sample
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
Application number
CN201811024483.3A
Other languages
Chinese (zh)
Inventor
吴刚
肖俊河
Current Assignee (The listed assignees may be inaccurate. Google has not performed a legal analysis and makes no representation or warranty as to the accuracy of the list.)
Shenzhen Mixlinker Network Co Ltd
Original Assignee
Shenzhen Mixlinker Network Co Ltd
Priority date (The priority date 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 date listed.)
Filing date
Publication date
Application filed by Shenzhen Mixlinker Network Co Ltd filed Critical Shenzhen Mixlinker Network Co Ltd
Priority to CN201811024483.3A priority Critical patent/CN109100975A/en
Publication of CN109100975A publication Critical patent/CN109100975A/en
Pending legal-status Critical Current

Links

Classifications

    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B19/00—Programme-control systems
    • G05B19/02—Programme-control systems electric
    • G05B19/04—Programme control other than numerical control, i.e. in sequence controllers or logic controllers
    • G05B19/042—Programme control other than numerical control, i.e. in sequence controllers or logic controllers using digital processors
    • G05B19/0428—Safety, monitoring
    • 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
    • G—PHYSICS
    • G05—CONTROLLING; REGULATING
    • G05B—CONTROL OR REGULATING SYSTEMS IN GENERAL; FUNCTIONAL ELEMENTS OF SUCH SYSTEMS; MONITORING OR TESTING ARRANGEMENTS FOR SUCH SYSTEMS OR ELEMENTS
    • G05B2219/00—Program-control systems
    • G05B2219/20—Pc systems
    • G05B2219/26—Pc applications
    • G05B2219/2609—Process control

Landscapes

  • Engineering & Computer Science (AREA)
  • Physics & Mathematics (AREA)
  • General Physics & Mathematics (AREA)
  • Theoretical Computer Science (AREA)
  • Data Mining & Analysis (AREA)
  • Evolutionary Computation (AREA)
  • Artificial Intelligence (AREA)
  • Biomedical Technology (AREA)
  • Biophysics (AREA)
  • Computational Linguistics (AREA)
  • Health & Medical Sciences (AREA)
  • Life Sciences & Earth Sciences (AREA)
  • General Health & Medical Sciences (AREA)
  • Molecular Biology (AREA)
  • Computing Systems (AREA)
  • General Engineering & Computer Science (AREA)
  • Mathematical Physics (AREA)
  • Software Systems (AREA)
  • Automation & Control Theory (AREA)
  • Feedback Control In General (AREA)

Abstract

The embodiment of the present application discloses a kind of parameter optimization method and system, is adjusted for the controllable parameter to equipment.The embodiment of the present application method includes: to obtain sample pair, and the sample is to including controllable parameter and object element；Model neural network based is constructed, which includes the relationship between the controllable parameter and the object element；Using the sample to the training model；Obtain expectation target element；The constraint condition of the controllable parameter is obtained, which defines the feasible zone of the controllable parameter；The controllable parameter is adjusted using the model and the value range, so that the object element is close to the expectation target element.

Description

A kind of parameter optimization method and system
Technical field
This application involves industrial control field more particularly to a kind of parameter optimization method and systems.
Background technique
With the development of industrial technology of Internet of things, all trades and professions start with internet of things equipment building Internet of Things profession service Platform is to realize the intelligent operation of various information.Be connected to Internet of Things equipment such as boiler, air compressor, heat pump can more than Equipment trip information is passed so that related technical personnel check and control.
The essence of parameter optimization is desirable to adjust certain some controllable parameter, and realization can not be straight for other Connect the estimated adjustment of the parameter of control.By taking boiler plant as an example, water instantaneous flow, furnace pressure, oxygen content, water level etc. are controllable Parameter processed may be implemented to fire box temperature by adjusting these controllable parameters, burn what output state etc. not directly controlled Parameter is adjusted.The operating status of equipment can't be carried out fully according to the idea of designer and user, due to controllable There is the relationship of interaction simultaneously between parameter, reach if attempting only to improve efficiency of combustion by adjusting the mode of oxygen content It is optimal, then it may cause the variation of other parameters so that it cannot reach desired efficiency of combustion.
Tradition mainly debugs the method that device parameter optimizes manually by technical staff, and technical staff first comments Estimate controllable parameter influence caused by object element, then applied in specific production process, assesses practical mesh Element is marked, is looped back and forth like this, until reaching desired object element.This kind of method time-consuming is too long, it is sometimes desirable to several weeks The even several months can be only achieved optimization aim.
Summary of the invention
The embodiment of the present application provides a kind of parameter optimization method and system, enables a device to reach optimum operation shape State.
In a first aspect, the embodiment of the present application provides a kind of method of parameter optimization, this method comprises:
The history log of equipment to be optimized is obtained, and therefrom obtains multiple groups sample pair, the sample is to by controllable ginseng Several and object element composition, the quantity of sample pair is sufficiently large, and the multiple groups sample is to being divided into training group and validation group again；
Initial model neural network based is constructed, the connection weight between neuron is defined in the model, can control Relationship between parameter and object element；
Using sample centering controllable parameter as the input of neural network model, object element is as neural network model Desired output, the training neural network model；
According to the practical operation situation of equipment, the value of setting expectation object element；
According to the relationship between the controllable parameter of equipment, the adjustable range of controllable parameter obtains these controllable ginsengs Several feasible zones；
Using the feasible zone of neural network model and controllable parameter after training, the value of controllable parameter is adjusted, is made Input of the controllable parameter as the neural network model after training is obtained, output can be close to the value of expectation target element.
According in a first aspect, in the first embodiment of the embodiment of the present application first aspect, which includes The numerical value of equipment runtime parameter, the parameter can be detected to obtain by the sensor in equipment, such as temperature sensor, pressure sensor Deng.
According in a first aspect, the building is based on nerve net in second of embodiment of the embodiment of the present application first aspect The initial model of network includes:
Input of the controllable parameter as input layer, output of the object element as output layer neuron, this is defeated Enter between layer and the output layer neuron there are hidden layer, between adjacent layer to connect entirely, it is connectionless between every layer of neuron；
The quantity of hidden layer neuron is determined according to the quantity of input layer and output layer neuron.
According to second of embodiment of first aspect, in the third embodiment of the embodiment of the present application first aspect, Using the sample, to training, the model includes:
Using the controllable parameter of sample centering as the input of neural network, the reality output result of neural network is obtained；
The connection weight between adjacent layer neuron is adjusted, until the reality output result of neural network and sample centering The error of the value of object element is minimum.
According to the third embodiment of first aspect, in the 4th kind of embodiment of the embodiment of the present application first aspect, Connection weight between the adjustment neuron includes:
Adjust the connection weight between the input layer and the hidden layer neuron；
Adjust the connection weight between the hidden layer neuron and the output layer neuron.
According in a first aspect, this is using the model and is somebody's turn to do in the 5th kind of embodiment of the embodiment of the present application first aspect Value range adjusts the controllable parameter
Determine the adjustment direction and adjusting step of controllable parameter；
The controllable parameter is adjusted according to the adjustment direction and the adjusting step, so that the output of neural network gradually approaches The value of the expectation target element.
Second aspect, the embodiment of the present application provide a kind of Parameter Optimization System, which includes:
First acquisition unit, for obtaining sample pair, the sample is to including controllable parameter and object element；
Modeling unit, for constructing model neural network based, which includes the controllable parameter and the target element Relationship between element；
Training unit, for utilizing the sample to the training model；
Second acquisition unit, for obtaining expectation target element；
Third acquiring unit, for obtaining the constraint condition of the controllable parameter, it is controllable which defines this The feasible zone of parameter；
Adjustment unit, for adjusting the controllable parameter using the model and the value range, so that the object element connects It is bordering on the expectation target element.
According to second aspect, in the first embodiment of the embodiment of the present application second aspect, which includes:
Subelement is inputted, obtains output result to the model for inputting the controllable parameter；
The first adjustment subelement, for adjusting the connection weight between neuron, until the output result and the target element Plain error is minimum.
According to the first embodiment of second aspect, in second of embodiment of the embodiment of the present application second aspect, The first adjustment subelement includes:
The first adjustment module, for adjusting the connection weight between the input layer and the hidden layer neuron；
Second adjustment module, for adjusting the connection weight between the hidden layer neuron and the output layer neuron.
According to second aspect, in the third embodiment of the embodiment of the present application second aspect, which includes:
Subelement is determined, for determining adjustment direction and adjusting step；
Second adjustment subelement, for adjusting the controllable parameter according to the adjustment direction and the adjusting step, so that should The object element of model output gradually approaches the expectation target element.
The third aspect, the embodiment of the present application provide a kind of Parameter Optimization System, and the Parameter Optimization System includes: processing Device and memory are stored with the instruction of parameter optimization described in aforementioned first aspect in the memory, when its on computers When operation, so that the step of computer executes the optimization of timing parameters described in aforementioned first aspect.
Fourth aspect, the embodiment of the present application provides a kind of computer readable storage medium, including instruction, when it is being calculated When being run on machine, so that computer executes method as described in relation to the first aspect.
As can be seen from the above technical solutions, the embodiment of the present application has the advantage that
The embodiment of the present application obtains the relationship between controllable parameter and object element by building neural network model, and Pass through controllable parameter corresponding to the Relation acquisition expectation target element.
Detailed description of the invention
Fig. 1 is a kind of BP neural network structural schematic diagram provided by the embodiments of the present application；
Fig. 2 is one embodiment schematic diagram of parameter optimization method provided by the embodiments of the present application；
Fig. 3 is one embodiment schematic diagram of Parameter Optimization System provided by the embodiments of the present application；
Fig. 4 is another embodiment schematic diagram of Parameter Optimization System provided by the embodiments of the present application.
Specific embodiment
The embodiment of the present application provides a kind of parameter optimization method and system, for making equipment by adjusting controllable parameter The object element of output reaches expectation target element, and the embodiment of the present application also provides corresponding Parameter Optimization System and computers Readable storage medium storing program for executing is described in detail separately below.
The structure of BP (back propagation, error back propagation) neural network is as shown in Figure 1, BP neural network is A kind of neural network network with three layers or more, including input layer (input layer), hidden layer (hidden layer) With output layer (output layer), the neuron between adjacent layer realizes full connection, and connectionless between every layer of neuron. After one section of learning sample is supplied to neural network, the activation value X of neuron1, X2…XMFrom input layer through each hidden layer to output Es-region propagations obtain the input response of network in each neuron of output layer.Later according to reduce desired output and reality output it Between the direction of error eventually passed back to from output layer by the connection weight between each intermediate each neuron of hidden layer layer-by-layer correction Input layer, corrects with this error back propagation and constantly carries out, and network also constantly rises the accuracy that input pattern responds.
One embodiment schematic diagram of parameter optimization method provided by the embodiments of the present application is as shown in Fig. 2, it mainly includes Six steps are embodied as follows:
201, sample pair is obtained.
Sample pair is obtained by device history log, which includes controllable parameter and object element.It is right In the representation of above-mentioned controllable parameter and object element, exist all in the form of parameter in physical device, in the application It is this kind of using controllable parameter as object element is adjusted in order to indicate the causality of control in embodiment and subsequent embodiment The means of parameter.The quantity of sample pair should be sufficiently large, and particular number should be according to the regulation gear of device parameter and controllable ginseng The quantity with object element is counted to choose, the multiple groups sample correspondence of selection is divided into training group and validation group, so as in training nerve Inspection can be carried out to its network fitness after network.
The parameter optimization method that the embodiment of the present application proposes is suitable for different industrial equipments, such as boiler plant, air Compressor apparatus, generating set etc..It can include: boiler pressure, boiler water with collected operating parameter by taking boiler plant as an example The various parameters information such as position, oxygen content, stop frequency, frequency converter output frequency, fire box temperature, in order to reach final control boiler The purpose of temperature, can choose boiler pressure, boiler water level, oxygen content, stop frequency, frequency converter output frequency is controllable ginseng Number, fire box temperature is object element, by adjusting controllable parameter to realize the control to object element, boiler plant parameter list As shown in table 1, it is notable that the value of parameters can be not the value of actual parameter in table, but carry out table Numerical value after sign extraction, is usually indicated with the value between 0-1, such as fire box temperature then may be used within the scope of 0-1000 DEG C 200 DEG C in sample data are expressed as 0.2 input neural network model.
Table 1
202, neural network model is constructed.
The operating status of industrial equipment is considered as a nonlinearity system about time series, therefore can use One three layers of neural network simulates the parameter of equipment.
In the embodiment of the present application, BP neural network as shown in Figure 1 is constructed, using controllable parameter as input, with mesh Element is marked as output.Wherein the quantity of input layer is the quantity of one group of sample centering controllable parameter, output layer mind Quantity through member is the quantity of one group of sample centering object element, if controlling an object element by multiple controllable parameters, Then the neuronal quantity of output layer is 1.
Selection for hidden layer neuron quantity is a relatively complicated problem, may when its quantity is very few The performance of network is influenced, the learning time that will lead to network when excessive is too long, and network fault tolerance performance is poor.If hidden layer neuron Quantity is m, and the quantity of input layer is n, and output layer neuron quantity is k, the neuron number for the hidden layer generally chosen Amount m should meetWherein constant or m=log of the α between 1-102N, orThese three Any one of condition.In the embodiment of the present application, to reduce possible calculating pressure, selection m=max { n, k } herein+ The m value of 1, i.e. m=n+1 are to meet the value of the neuronal quantity of minimum hidden layer of neural net model establishing condition, input sample into Row training, if in training process, when reaching maximum number of iterations, the capability of fitting of the neural network is still weaker, then gradually Increase the value of m, until its is eligible.
Input and output for hidden layer, generally have:
WhereinThe expression second layer, i.e. the input of hidden layer,For the output of hidden layer, k is hidden layer nerve The quantity of member,For the output of input layer,Indicate that j-th of neuron in first layer is directed toward i-th of mind in the second layer Connection weight through member.Wherein function f (x) is excitation function, takes Sigmoid function:
Its codomain is between (0,1), according to the method described above, successively establishes the relationship between each layer neuron, that is, completes The initialization of neural network models.
203, training neural network model.
Using the sample data of training group in the multiple groups sample obtained in step 201 to the BP nerve constructed in step 202 Netinit model is trained, by the real output value and desired output in neural network, the i.e. target of sample centering The value of element compares, and corrects the connection weight between neuron according to comparing result, until the reality output of neural network is most It is possibly close to the object element of sample centering.Its principle is, when the input of network is X=(x1,x2,…,xn) when, network Reality output Y=(y1,y2,…,yn), the desired output D=(d of network1,d2,…,dn).Define learning function, i.e. reality output Mean square deviation between desired output:
BP algorithm passes throughThe connection weight between neuron is adjusted, so that above-mentioned formula Value it is minimum, wherein η is learning rate.The non-linear relation between controllable parameter and object element is obtained with this, is denoted as: Y ≈ FBP(X1,X2,…XN)
204, expectation target element is obtained.
Object element is set and it is expected the numerical value that reaches, by taking boiler plant as an example, parameter when according to equipment actual motion, if Setting the data that fire box temperature expectation reaches is 800 DEG C, and the method for setting expectation object element can be going through according to equipment operation Records of the Historian record, is also possible to technical staff and is configured according to actual production demand.
205, constraint condition is obtained.
Based on the interaction relationship between device history log or controllable parameter, taking for controllable parameter is obtained The relationship between controllable parameter is quantitatively described in the feasible zone of value.
Remember X=(X1,X2,…,XN), then functional relation can be abbreviated as Y=F (X), for each controllable parameter Xi, There are reasonable ranges for its own, and such as the percentage of oxygen content, the height of boiler water level can all be limited, simultaneously by equipment There is also certain interaction relationship between these controllable parameters, these relationships are generally easier to determine, these are controllable The constraint condition G of parameter processed(X)It is denoted as:
In the embodiment of the present application, by the constraint condition of given controllable parameter, the feasible zone of controllable parameter is obtained, The controllable parameter obtained in step 206 is enabled to meet actual industrial production demand.
206, controllable parameter is adjusted.
Controllable parameter is adjusted in the case where meeting the constraint condition in step 205, so that inputting in controllable parameter To after the BP neural network model got in step 203, output can reach the expectation target element in step 204 Value.
In the embodiment of the present application, the purpose of parameter optimization is to keep the difference of realistic objective element and expectation target element most It is possible small.Therefore it can enableObtain following prioritization scheme:
s.t.X∈G(X)
WhereinIt is the expectation target element of object element Y.Finding the corresponding controllable parameter of expectation target element In the process, because the value of controllable parameter is the process gradually changed, then the expression formula of following Non-Linear Programming can be obtained:
Wherein X(0)The corresponding value of controllable parameter in equipment current operating conditions is taken, for above-mentioned optimization problem, to set Standby current operating conditions are origin, determine the direction of search and step-size in search, search for a series of point x(1),x(2),x(3)..., often The controllable parameter generated after an iteration as subsequent time equipment run when controllable parameter, make these point it is corresponding can Control parameter gradually approaches the point of optimal value, i.e. the corresponding controllable parameter of expectation object element, it is comprehensive obtain equipment about The prioritization scheme of the controllable parameter of time.
In the embodiment of the present application, neural network model also needs to utilize sample pair in step 201 after training in step 203 Validation group data verified.The corresponding controllable ginseng of expectation target element is obtained by way of establishing neural network model Number, improves the practicability of the embodiment of the present application.
Parameter optimization method in the present embodiment is described above, it is excellent to parameter provided by the embodiments of the present application below Change system is introduced.It is illustrated in figure 3 one embodiment signal of Parameter Optimization System 300 provided by the embodiments of the present application Figure, one embodiment of Parameter Optimization System 300 include:
First acquisition unit 301, for obtaining sample pair, the sample is to including controllable parameter and object element；
Modeling unit 302, for constructing model neural network based, which includes the controllable parameter and the target Relationship between element；
Training unit 303, for utilizing the sample to the training model；
Second acquisition unit 304, for obtaining expectation target element；
Third acquiring unit 305, for obtaining the constraint condition of the controllable parameter, it is controllable which defines this The feasible zone of parameter processed；
Adjustment unit 306, for adjusting the controllable parameter using the model and the value range, so that the object element Close to the expectation target element.
In the present embodiment, training unit 303 includes:
Subelement 3030 is inputted, obtains output result to the model for inputting the controllable parameter；
The first adjustment subelement 3031, for adjusting the connection weight between neuron, until the output result and the mesh It is minimum to mark element error.
In the present embodiment, the first adjustment subelement 3031 includes:
The first adjustment module 30310, for adjusting the connection weight between the input layer and the hidden layer neuron Weight；
Second adjustment module 30311, for adjusting the connection weight between the hidden layer neuron and the output layer neuron Weight.
In the present embodiment, adjustment unit 306 includes:
Subelement 3060 is determined, for determining adjustment direction and adjusting step；
Second adjustment subelement 3061 makes for adjusting the controllable parameter according to the adjustment direction and the adjusting step The object element for obtaining model output gradually approaches the expectation target element.
Fig. 4 is the structural schematic diagram of Parameter Optimization System 400 provided by the embodiments of the present application.The Parameter Optimization System 400 include processor 401, memory 402 and input and output (I/O) interface 403, and memory 402 may include read-only memory And random access memory, and operational order and data are provided to processor 401.The a part of of memory 402 can also include Nonvolatile RAM (NVRAM).
The operation of 401 Optimization about control parameter system 400 of processor, processor 401 can also be known as CPU.Memory 402 can To include read-only memory and random 4 access memory, and instruction and data is provided to processor 401.One of memory 402 Dividing can also include nonvolatile RAM (NVRAM).Each group of Parameter Optimization System 400 in specific application Part is coupled by bus system 404, and wherein bus system 404 can also include power supply in addition to including data/address bus Bus, control bus and status signal bus in addition etc..But for the sake of clear explanation, various buses are all designated as bus in figure System 404.
The method that above-mentioned the embodiment of the present application discloses can be applied in processor 401, or be realized by processor 401. Processor 401 may be a kind of IC chip, the processing capacity with signal.During realization, the above method it is each Step can be completed by the integrated logic circuit of the hardware in processor 401 or the instruction of software form.Above-mentioned processing Device 401 can be general processor, digital signal processor (DSP), specific integrated circuit (ASIC), ready-made programmable gate array (FPGA) either other programmable logic device, discrete gate or transistor logic, discrete hardware components.May be implemented or Person executes disclosed each method, step and logic diagram in the embodiment of the present application.General processor can be microprocessor or Person's processor is also possible to any conventional processor etc..The step of method in conjunction with disclosed in the embodiment of the present application, can be straight Connect and be presented as that hardware decoding processor executes completion, or in decoding processor hardware and software module combination executed At.Software module can be located at random access memory, and flash memory, read-only memory, programmable read only memory or electrically-erasable can In the storage medium of this fields such as programmable memory, register maturation.The storage medium is located at memory 402, and processor 401 is read Information in access to memory 402, in conjunction with the step of its hardware completion above method.
It is apparent to those skilled in the art that for convenience and simplicity of description, the system of foregoing description, The specific work process of device and unit, can refer to corresponding processes in the foregoing method embodiment, and details are not described herein.
In several embodiments provided herein, it should be understood that disclosed system, device and method can be with It realizes by another way.For example, system embodiment described above is only schematical, for example, the unit It divides, only a kind of logical function partition, there may be another division manner in actual implementation, such as multiple units or components It can be combined or can be integrated into another system, or some features can be ignored or not executed.Another point, it is shown or The mutual coupling, direct-coupling or communication connection discussed can be through some interfaces, the indirect coupling of system or unit It closes or communicates to connect, can be electrical property, mechanical or other forms.
The unit as illustrated by the separation member may or may not be physically separated, aobvious as unit The component shown may or may not be physical unit, it can and it is in one place, or may be distributed over multiple In network unit.It can select some or all of unit therein according to the actual needs to realize the mesh of this embodiment scheme 's.
It, can also be in addition, each functional unit in each embodiment of the application can integrate in one processing unit It is that each unit physically exists alone, can also be integrated in one unit with two or more units.Above-mentioned integrated list Member both can take the form of hardware realization, can also realize in the form of software functional units.
If the integrated unit is realized in the form of SFU software functional unit and sells or use as independent product When, it can store in a computer readable storage medium.Based on this understanding, the technical solution of the application is substantially The all or part of the part that contributes to existing technology or the technical solution can be in the form of software products in other words It embodies, which is stored in a storage medium, including some instructions are used so that a computer Equipment (can be personal computer, server or the network equipment etc.) executes the complete of each embodiment the method for the application Portion or part steps.And storage medium above-mentioned include: USB flash disk, mobile hard disk, read-only memory (read-only memory, ROM), random access memory (random access memory, RAM), magnetic or disk etc. are various can store program The medium of code.
The above, above embodiments are only to illustrate the technical solution of the application, rather than its limitations；Although referring to before Embodiment is stated the application is described in detail, those skilled in the art should understand that: it still can be to preceding Technical solution documented by each embodiment is stated to modify or equivalent replacement of some of the technical features；And these It modifies or replaces, the spirit and scope of each embodiment technical solution of the application that it does not separate the essence of the corresponding technical solution.

Claims (10)

1. a kind of parameter optimization method, which is characterized in that the described method includes:
Sample pair is obtained, the sample is to including controllable parameter and object element；
Model neural network based is constructed, the model includes the pass between the controllable parameter and the object element System；
Using the sample the training model；
Obtain expectation target element；
The constraint condition of the controllable parameter is obtained, the constraint condition defines the feasible zone of the controllable parameter；
The controllable parameter is adjusted using the model and the value range, so that the object element is close to the phase Hope object element.
2. the method according to claim 1, wherein the controllable parameter includes equipment runtime parameter, institute The numerical value for stating parameter can be detected to obtain by the sensor in equipment.
3. the method according to claim 1, wherein the building model neural network based includes:
Input of the controllable parameter as input layer, output of the object element as output layer neuron, There are hidden layers between the input layer and the output layer；
Determine the quantity of hidden layer neuron.
4. according to the method described in claim 3, it is characterized in that, described include: the training model using the sample
It inputs the controllable parameter and obtains output result to the model；
The connection weight between the neuron is adjusted, until the output result and the object element error are minimum.
5. according to the method described in claim 4, it is characterized in that, the connection weight packet adjusted between the neuron It includes:
Adjust the connection weight between the input layer and the hidden layer neuron；
Adjust the connection weight between the hidden layer neuron and the output layer neuron.
6. the method according to claim 1, wherein described adjust institute using the model and the value range Stating controllable parameter includes:
Determine adjustment direction and adjusting step；
The controllable parameter is adjusted according to the adjustment direction and the adjusting step, so that the mesh of model output Mark element gradually approaches the expectation target element.
7. a kind of Parameter Optimization System is applied to industrial control field characterized by comprising
First acquisition unit, for obtaining sample pair, the sample is to including controllable parameter and object element；
Modeling unit, for constructing model neural network based, the model includes the controllable parameter and the target Relationship between element；
Training unit, for utilizing the sample the training model；
Second acquisition unit, for obtaining expectation target element；
Third acquiring unit, for obtaining the constraint condition of the controllable parameter, the constraint condition defines described controllable The feasible zone of parameter processed；
Adjustment unit, for adjusting the controllable parameter using the model and the value range, so that the target element Element is close to the expectation target element.
8. Parameter Optimization System according to claim 7, which is characterized in that the training unit includes:
Subelement is inputted, obtains output result to the model for inputting the adjustment controllable parameter；
Subelement is adjusted, for adjusting the connection weight between neuron, until the output result and the object element miss It is poor minimum.
9. a kind of Parameter Optimization System, which is characterized in that the Parameter Optimization System includes: processor and memory, described to deposit The instruction of any parameter optimization of claim 1-6 is stored in reservoir, the processor is joined for executing in memory The step of counting the instruction of optimization, executing the parameter optimization method as described in claim 1-6 is any.
10. a kind of computer readable storage medium, which is characterized in that it is excellent to be stored with parameter in the computer readable storage medium The instruction of change, when run on a computer, so that computer executes any method of the claims 1-6.
CN201811024483.3A 2018-09-03 2018-09-03 A kind of parameter optimization method and system Pending CN109100975A (en)

Priority Applications (1)

Application Number Priority Date Filing Date Title
CN201811024483.3A CN109100975A (en) 2018-09-03 2018-09-03 A kind of parameter optimization method and system

Applications Claiming Priority (1)

Application Number Priority Date Filing Date Title
CN201811024483.3A CN109100975A (en) 2018-09-03 2018-09-03 A kind of parameter optimization method and system

Publications (1)

Publication Number Publication Date
CN109100975A true CN109100975A (en) 2018-12-28

Family

ID=64865029

Family Applications (1)

Application Number Title Priority Date Filing Date
CN201811024483.3A Pending CN109100975A (en) 2018-09-03 2018-09-03 A kind of parameter optimization method and system

Country Status (1)

Country Link
CN (1) CN109100975A (en)

Cited By (7)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110059892A (en) * 2019-04-29 2019-07-26 贾原琪 A kind of more physical quantitys monitor the method to carry out industrial production optimization
CN111900996A (en) * 2020-08-06 2020-11-06 山东超越数控电子股份有限公司 Method, system, equipment and medium for determining CRC32 algorithm parameters
CN112691783A (en) * 2020-12-09 2021-04-23 华润电力技术研究院有限公司 Pulverized coal boiler pulverizing system regulation and control method, device, equipment and storage medium
CN112836875A (en) * 2021-02-02 2021-05-25 朗坤智慧科技股份有限公司 Equipment regulation and control method and system based on time sequence domain and network side server
CN113708781A (en) * 2021-08-13 2021-11-26 Oppo广东移动通信有限公司 Radio frequency gain control method, device and communication equipment
CN113777965A (en) * 2020-05-21 2021-12-10 广东博智林机器人有限公司 Spraying quality control method and device, computer equipment and storage medium
CN114244710A (en) * 2021-12-31 2022-03-25 中国电信股份有限公司 Network element parameter adjusting method and device and electronic equipment

Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102626557A (en) * 2012-04-13 2012-08-08 长春工业大学 Molecular distillation process parameter optimizing method based on GA-BP (Genetic Algorithm-Back Propagation) algorithm
CN102831269A (en) * 2012-08-16 2012-12-19 内蒙古科技大学 Method for determining technological parameters in flow industrial process
CN103324862A (en) * 2013-07-11 2013-09-25 中国石油大学(华东) Coal-fired boiler optimization method based on improved neural network and genetic algorithm
CN107463747A (en) * 2017-08-03 2017-12-12 重庆工业职业技术学院 The application of neural network model and construction method in the heat treatment of light-alloy parts
US20180096244A1 (en) * 2016-09-30 2018-04-05 Sony Interactive Entertainment Inc. Method and system for classifying virtual reality (vr) content based on modeled discomfort of a user
CN108388702A (en) * 2018-01-30 2018-08-10 河南工程学院 Engineering ceramics electrical discharge machining effect prediction method based on PSO neural networks

Patent Citations (6)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN102626557A (en) * 2012-04-13 2012-08-08 长春工业大学 Molecular distillation process parameter optimizing method based on GA-BP (Genetic Algorithm-Back Propagation) algorithm
CN102831269A (en) * 2012-08-16 2012-12-19 内蒙古科技大学 Method for determining technological parameters in flow industrial process
CN103324862A (en) * 2013-07-11 2013-09-25 中国石油大学(华东) Coal-fired boiler optimization method based on improved neural network and genetic algorithm
US20180096244A1 (en) * 2016-09-30 2018-04-05 Sony Interactive Entertainment Inc. Method and system for classifying virtual reality (vr) content based on modeled discomfort of a user
CN107463747A (en) * 2017-08-03 2017-12-12 重庆工业职业技术学院 The application of neural network model and construction method in the heat treatment of light-alloy parts
CN108388702A (en) * 2018-01-30 2018-08-10 河南工程学院 Engineering ceramics electrical discharge machining effect prediction method based on PSO neural networks

Cited By (12)

* Cited by examiner, † Cited by third party
Publication number Priority date Publication date Assignee Title
CN110059892A (en) * 2019-04-29 2019-07-26 贾原琪 A kind of more physical quantitys monitor the method to carry out industrial production optimization
CN113777965A (en) * 2020-05-21 2021-12-10 广东博智林机器人有限公司 Spraying quality control method and device, computer equipment and storage medium
CN113777965B (en) * 2020-05-21 2023-07-18 广东博智林机器人有限公司 Spray quality control method, spray quality control device, computer equipment and storage medium
CN111900996A (en) * 2020-08-06 2020-11-06 山东超越数控电子股份有限公司 Method, system, equipment and medium for determining CRC32 algorithm parameters
CN111900996B (en) * 2020-08-06 2024-03-15 超越科技股份有限公司 Determination method, system, equipment and medium for CRC32 algorithm parameters
CN112691783A (en) * 2020-12-09 2021-04-23 华润电力技术研究院有限公司 Pulverized coal boiler pulverizing system regulation and control method, device, equipment and storage medium
CN112836875A (en) * 2021-02-02 2021-05-25 朗坤智慧科技股份有限公司 Equipment regulation and control method and system based on time sequence domain and network side server
CN113708781A (en) * 2021-08-13 2021-11-26 Oppo广东移动通信有限公司 Radio frequency gain control method, device and communication equipment
WO2023016231A1 (en) * 2021-08-13 2023-02-16 Oppo广东移动通信有限公司 Radio frequency gain control method and apparatus, and communication device
CN113708781B (en) * 2021-08-13 2023-02-17 Oppo广东移动通信有限公司 Radio frequency gain control method, device and communication equipment
CN114244710A (en) * 2021-12-31 2022-03-25 中国电信股份有限公司 Network element parameter adjusting method and device and electronic equipment
CN114244710B (en) * 2021-12-31 2024-06-28 中国电信股份有限公司 Network element parameter adjustment method and device and electronic equipment

Similar Documents

Publication Publication Date Title
CN109100975A (en) A kind of parameter optimization method and system
US11915142B2 (en) Creating equipment control sequences from constraint data
US20200175435A1 (en) System and method for optimizing combustion of boiler
CN116627027B (en) Optimal robustness control method based on improved PID
CN112052992A (en) Building engineering project progress prediction system and method based on deep learning
CN110826774B (en) Bus load prediction method and device, computer equipment and storage medium
CN105556401B (en) For area of computer aided control and/or regulate the method for technological system
WO2019176496A1 (en) Control device, control system, control method, and control program
CN101059846A (en) Methods and systems for predictive modeling using a committee of models
CN117031950A (en) Modeling method and device for deep peak-shaving thermal power generating unit control system
CN109034452A (en) Energy consumption prediction technique for auto-parts manufacturing enterprise
Ellis Machine learning enhanced grey-box modeling for building thermal modeling
CN102955429A (en) Process for adaptive modeling of performance degradation
Mitterhofer et al. Semantics for assembling modular components in a scalable building performance simulation
US20210248442A1 (en) Computing device and method using a neural network to predict values of an input variable of a software
CN111279276B (en) Randomization reinforcement learning for controlling complex systems
CN108319146B (en) Discrete particle swarm training-based method for radial basis function neural network
Stoffel et al. Combining data-driven and physics-based process models for hybrid model predictive control of building energy systems
CN114118611A (en) Wind power prediction method and system based on prediction algorithm pool
CN103558762B (en) The implementation method of the immune genetic PID controller based on graphical configuration technology
CN109670636B (en) Optimization method and device of energy scheduling model for universal energy station
US20200348631A1 (en) Method for the computer-aided control of a technical system, more particularly a power generation installation
Senyard et al. Software engineering methods for neural networks
Hadiyat et al. Comparing statistical feature and artificial neural networks for control chart pattern recognition: a case study
US20240013096A1 (en) Dual-model machine learning for process control and rules controller for manufacturing equipment

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
RJ01 Rejection of invention patent application after publication

Application publication date: 20181228