CN110263410A - Machine utilization interval prediction method and computer readable storage medium based on fuzzy neural network - Google Patents
Machine utilization interval prediction method and computer readable storage medium based on fuzzy neural network Download PDFInfo
- Publication number
- CN110263410A CN110263410A CN201910514109.XA CN201910514109A CN110263410A CN 110263410 A CN110263410 A CN 110263410A CN 201910514109 A CN201910514109 A CN 201910514109A CN 110263410 A CN110263410 A CN 110263410A
- Authority
- CN
- China
- Prior art keywords
- rule
- fuzzy
- neural network
- ith
- algorithm
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 50
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 29
- 238000003860 storage Methods 0.000 title claims abstract description 8
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 37
- 238000004364 calculation method Methods 0.000 claims description 12
- 238000013138 pruning Methods 0.000 claims description 8
- 238000004590 computer program Methods 0.000 claims description 4
- 238000002474 experimental method Methods 0.000 claims description 4
- 230000004931 aggregating effect Effects 0.000 claims description 3
- 238000005315 distribution function Methods 0.000 claims description 3
- 239000011159 matrix material Substances 0.000 claims description 3
- 238000010998 test method Methods 0.000 claims description 3
- 238000009966 trimming Methods 0.000 claims description 2
- 238000011156 evaluation Methods 0.000 abstract 1
- 238000004519 manufacturing process Methods 0.000 description 6
- 108090000672 Annexin A5 Proteins 0.000 description 3
- 230000008859 change Effects 0.000 description 3
- 238000010586 diagram Methods 0.000 description 3
- 238000013473 artificial intelligence Methods 0.000 description 2
- 230000008901 benefit Effects 0.000 description 2
- 239000004020 conductor Substances 0.000 description 2
- 238000012549 training Methods 0.000 description 2
- 238000004458 analytical method Methods 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 239000003245 coal Substances 0.000 description 1
- 238000002485 combustion reaction Methods 0.000 description 1
- 239000002131 composite material Substances 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 238000009472 formulation Methods 0.000 description 1
- 239000002803 fossil fuel Substances 0.000 description 1
- 230000004927 fusion Effects 0.000 description 1
- 238000009776 industrial production Methods 0.000 description 1
- 210000002364 input neuron Anatomy 0.000 description 1
- 238000013507 mapping Methods 0.000 description 1
- 230000007246 mechanism Effects 0.000 description 1
- 238000005272 metallurgy Methods 0.000 description 1
- 239000000203 mixture Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 229940037201 oris Drugs 0.000 description 1
- 210000004205 output neuron Anatomy 0.000 description 1
- 239000003208 petroleum Substances 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 230000004044 response Effects 0.000 description 1
- 238000012360 testing method Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F30/00—Computer-aided design [CAD]
- G06F30/20—Design optimisation, verification or simulation
-
- 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
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Evolutionary Computation (AREA)
- General Engineering & Computer Science (AREA)
- Artificial Intelligence (AREA)
- Computational Linguistics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Computer Hardware Design (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Geometry (AREA)
- Data Mining & Analysis (AREA)
- General Health & Medical Sciences (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The present invention provides a kind of machine utilization interval prediction method and computer readable storage medium based on fuzzy neural network, machine utilization interval prediction method, step includes: S1 initialization rule base, fuzzy rule is generated based on the potential of iteration density estimation method computation rule, and according to the first rule-based algorithm；S2 calculates the similarity of fuzzy rule based on subordinating degree function analogue method, to merge fuzzy rule according to third rule-based algorithm；S3 is based on fuzzy weighted values recursive least squares algorithm, to fuzzy rule consequent parameter online updating；S4 is based on Bootstrap method construct forecast interval.After adopting the above technical scheme, can provide intelligence, the variation of quick equipment internal loading amplitude estimate and confidence evaluation.
Description
Technical Field
The invention relates to the field of intelligent control, in particular to a fuzzy neural network-based equipment load interval prediction method and a computer-readable storage medium.
Background
The air compressor is the main power energy source in the industrial production processes of metallurgy, industrial manufacture, biological pharmacy and the like. With the shortage of primary energy sources such as coal, petroleum and the like, the reasonable scheduling of the air compressor group can not only improve the energy-saving and consumption-reducing level of enterprises, increase the economic benefits of the enterprises, but also reduce the pollution of fossil fuel combustion to the environment.
In order to realize the optimized dispatching of the air compressor unit, production personnel need to master the change trend of the load of the air compressor group in time. Therefore, under the condition of accurately predicting the load flow of the air compressor group in time, on one hand, production personnel can be assisted to make a safe and economic scheduling scheme, and the air consumption requirements of different production users are met; on the other hand, the energy saving performance and the economical efficiency of the air compressor group can be improved, and a foundation support is provided for production safety, energy saving and consumption reduction.
Meanwhile, with the rapid development of artificial intelligence technology, artificial intelligence needs to be introduced in the traditional industry to assist in solving the production problem. Thus, neural network algorithms can be introduced to achieve a prediction of the device load. The fuzzy neural network is a product combining the knowledge expression capability of fuzzy logic reasoning and the self-learning capability of the neural network, and is widely used for solving the problems of fuzzy regression, fuzzy pedigree analysis, fuzzy controller design, fuzzy expert system and the like.
Therefore, the invention provides a novel device load interval prediction method based on the fuzzy neural network, which can accurately estimate the load amplitude change in the device and assist the formulation of a field operation scheme.
Disclosure of Invention
In order to overcome the above technical drawbacks, an object of the present invention is to provide a method for predicting a load interval of a device based on a fuzzy neural network and a computer readable storage medium, which can provide intelligent and fast prediction of load amplitude variation in the device and measure reliability.
The invention discloses a device load interval prediction method based on a fuzzy neural network, which comprises the following steps:
s1 initializing a rule base, calculating the potential of the rule based on an iterative density estimation method, and generating a fuzzy rule according to a first rule algorithm;
s2 calculating the similarity of the fuzzy rules based on the membership function similarity method, so as to fuse the fuzzy rules according to a third rule algorithm;
s3, updating the parameters of the fuzzy rule postware on line based on the fuzzy weight recursive least square algorithm;
s4 constructs a prediction interval based on the Bootstrap method.
Preferably, the steps further comprise:
s1.1, based on a second rule algorithm, trimming and recalling fuzzy rules.
The fuzzy neural network-based device load interval prediction method of claim 2, wherein said second rule algorithm step comprises:
s1.1.1, calculating the closeness degree of the output data of the ith rule in the fuzzy rule and the real output data based on the following formula:
wherein d (N) represents the Nth real output data, yi(N) output data representing the ith rule;indicating the proximity of the nth output data of the ith rule to the real output.
S1.1.2 calculating the Life Strength P of the ith rule based on the following formulai(N)：
Wherein, the forgetting factor zeta belongs to (0, 1);
when the life intensity PiWhen (N) < ═ thresP, pruning the ith rule from the fuzzy rule, wherein thresP represents a rule pruning threshold value;
s1.1.3 according to the formula:it is determined whether the ith rule is recalled, wherein,κ represents the number of pruned rules and the number of untrimmed rules in the rule base, respectively.
Preferably, the step of calculating the potential of the rule based on the iterative density estimation method comprises:
the first-order Taylor expansion is carried out by utilizing the Gaussian membership function to obtain:
wherein Represents the nth sample;
the newly generated sample density is calculated based on the following formula:
wherein
Preferably, wherein the first rule algorithm step comprises:
s1.2 according to the conditions:orAt least one of, judging to generate fuzzy rules;
s1.3 calculation formula: cnew＝xN，ΩnewRand (u) to initialize the parameters of the newly generated fuzzy rule, wherein
CnewDenotes the center, Ω, of the newly generated fuzzy rulenewA back-piece parameter representing the newly generated fuzzy rule.
Preferably, the membership function similarity calculation comprises the following steps:
s2.1 calculation formula:
where cwin, j and ci, j indicate that the winning rule win and the ith rule are centered in the jth dimension,andrespectively representing the variances of the winning rule win and the ith rule in the jth dimension;
s2.2, taking the maximum membership degree of the intersection point coordinates of at least two membership degree functions as the similarity of the subsets, calculating the similarity of all the subsets of the front piece in the fuzzy rule according to the following formula, and aggregating to obtain the overlapping degree of the winning rule win and the rule i:
s2.3 according to Si(win, i) > 0 and overlapwin，iSelecting a rule with rule similarity greater than a set value threrespap;
and S2.4, if the inequality condition of the step S2.3 is met, fusing the fuzzy rule by using a third rule algorithm.
The fuzzy neural network-based device load interval prediction method of claim 1, wherein said third rule algorithm calculation step comprises:
s2.5 fuses the fuzzy rules according to the following formula:
respectively acquiring the fused rule centers CmergedVariance σ2 mergedAnd a back-part parameter omegamerged，
wherein Andrespectively representing the number of samples in the winning rule and the ith fuzzy rule;andrespectively representing the variances of the winning rule and the ith fuzzy rule; omegawin and ΩiThe back-piece parameters of the winning rule win and the ith fuzzy rule are respectively represented.
Preferably, the fuzzy weight recursive least squares algorithm calculating step comprises:
s3.1 updating the back-piece parameters based on the following formula:
wherein ,i is an identity matrix of a back-part parameter, A represents a forgetting factor, omegaNRepresenting the nth iteration of the back-piece parameter.
Preferably, the step of constructing the prediction interval based on the boottrap method comprises:
s4.1 constructs a prediction interval based on:
wherein , represents the average value of the predicted output after theta independent repeated experiments; t is tθ/2(theta) is the (theta/2) quantile of the t-distribution function with theta degrees of freedom, sigma2 noisyRepresents the variance of the noise estimate in the data using the Gamma Test method.
The invention also discloses a computer readable storage medium on which a computer program is stored, which, when executed by a processor, implements a device coincidence interval prediction method as described above.
After the technical scheme is adopted, compared with the prior art, the method has the following beneficial effects:
1. can be equipment such as an air compressor and the like; providing an accurate estimate of load amplitude variation and a corresponding confidence measure;
2. the equipment application site can specify a safe and economic scheduling scheme according to the prediction interval result, so that the resource utilization rate of the equipment is improved;
3. by utilizing the fuzzy neural network, a rapid, efficient and low-cost estimation means can be realized, and the intelligent level of the traditional industry is greatly improved.
Drawings
FIG. 1 is a schematic diagram of a topology of a fuzzy neural network consistent with the present invention;
FIG. 2 is a logic diagram illustrating a method for predicting a device load interval in accordance with a preferred embodiment of the present invention;
fig. 3 is a flow chart illustrating a method for predicting a device load interval according to a preferred embodiment of the present invention.
Detailed Description
The advantages of the invention are further illustrated in the following description of specific embodiments in conjunction with the accompanying drawings.
Reference will now be made in detail to the exemplary embodiments, examples of which are illustrated in the accompanying drawings. When the following description refers to the accompanying drawings, like numbers in different drawings represent the same or similar elements unless otherwise indicated. The implementations described in the exemplary embodiments below are not intended to represent all implementations consistent with the present disclosure. Rather, they are merely examples of apparatus and methods consistent with certain aspects of the present disclosure, as detailed in the appended claims.
The terminology used in the present disclosure is for the purpose of describing particular embodiments only and is not intended to be limiting of the disclosure. As used in this disclosure and the appended claims, the singular forms "a," "an," and "the" are intended to include the plural forms as well, unless the context clearly indicates otherwise. It should also be understood that the term "and/or" as used herein refers to and encompasses any and all possible combinations of one or more of the associated listed items.
It is to be understood that although the terms first, second, third, etc. may be used herein to describe various information, such information should not be limited to these terms. These terms are only used to distinguish one type of information from another. For example, first information may also be referred to as second information, and similarly, second information may also be referred to as first information, without departing from the scope of the present disclosure. The word "if" as used herein may be interpreted as "at … …" or "when … …" or "in response to a determination", depending on the context.
Referring to fig. 1, a schematic diagram of a fuzzy neural network topology according to the present invention is shown. Between input neurons and output neurons, through mapping of a feedforward neural network, for example, a first layer gives input data semanteme, a second layer normalizes the credibility of an ith rule, and the output of a final layer self-adaptive fuzzy neural network can be represented as a linear combination of post-piece parameters (when the pre-piece parameters are fixed, the post-piece parameters identified by an LSE method are optimal). Fig. 2 shows the basic logic relied on by the fuzzy neural network, after obtaining a newly input training sample, firstly, whether the rule base is empty needs to be judged, then, the fuzzy rule parameters are initialized or the fuzzy rules are pruned and then recalled to judge whether the formed new fuzzy rules meet the generation conditions, if the conditions are met, the rule parameters can be initialized, and if the conditions are not met, the fuzzy rules are pruned and fused in multiple steps, and the fuzzy rules are adaptively changed by updating the background parameters.
Referring to fig. 3, based on the basic topology and logic, the method for predicting the device load interval based on the fuzzy neural network in the present invention includes the steps of:
s1: initializing a rule base, for example, using a first piece of data in a training data set or a historical sample as initialization basic data, calculating the potential of the rule based on an iterative density estimation method, and generating a fuzzy rule according to a first rule algorithm, wherein the first rule algorithm can be a judgment condition for judging whether the new rule meets a generation condition and whether the new rule can be generated, and initializing a front-piece parameter and a back-piece parameter of the new rule when the generation condition is met;
s2: calculating the similarity of the fuzzy rules based on a membership function similarity method, fusing the fuzzy rules according to a third rule algorithm, and realizing the compromise of interpretability and compactness of the rule base by fusing the similar fuzzy rules;
s3: updating the parameters of the fuzzy rule back-part on line based on a fuzzy weight recursive least square algorithm;
s4: and constructing a prediction interval based on a Bootstrap method.
In a preferred embodiment, the method for predicting the device load interval further includes S1.1: based on the second planning algorithm, pruning and recalling are carried out on the fuzzy rule, and catastrophic forgetting of the pruning rule can be avoided through a corresponding pruning and recalling mechanism. Specifically, step S1.1 comprises:
s1.1.1: calculating the closeness degree of the output data of the ith rule in the fuzzy rule and the real output data based on the following formula:
whereinD (N) represents the Nth actual output data of all data, yi(N) output data representing the ith rule:representing the closeness of the Nth output data of the ith rule to the real output;
s1.1.2 calculating the Life Strength P of the ith rule based on the following formulai(N)：
Where ζ represents a forgetting factor and ζ ∈ (0, 1); it is understood that smaller ζ decays faster.
When the life intensity PiWhen (N) < ═ thresP, pruning the ith rule from the fuzzy rule, wherein thresP represents a rule pruning threshold value;
s1.1.3 according to the formula:it is determined whether the ith rule is recalled, wherein,κ represents the number of pruned rules and the number of untrimmed rules in the rule base, respectively.
In another preferred or alternative embodiment, the step of calculating the potential of the rule based on the iterative density estimation method comprises:
the first-order Taylor expansion is carried out by utilizing the Gaussian membership function to obtain:
wherein Representing the nth sample.
Then, the newly generated sample density is calculated based on the following formula:
in the formula
Further, after the density of the new generated sample is obtained, whether a new fuzzy rule is generated or not is judged according to the following conditions, namely, the first rule algorithm comprises:
s1.2 according to the conditions:orIs judged to generate the fuzzy rule, it is understood that when the condition is satisfiedWhen the rule is generated, a plurality of samples exist around the newly generated node according to the new fuzzy rule, and the generation of the new node can improve the generalization capability of the rule; when the condition is satisfiedThen, it means that the newly generated node can be complementedFilling the area which is not covered by the previous fuzzy rule, thereby being capable of adapting to the change of the external environment and improving the generalization performance of the newly generated fuzzy rule;
s1.3 calculation formula: cnew＝xN，ΩnewRand (u) to initialize the parameters of the newly generated fuzzy rule, where CnewDenotes the center, Ω, of the newly generated fuzzy rulenewA back-piece parameter representing the newly generated fuzzy rule.
Through the configuration, the accuracy of the rules in the neural network is continuously optimized and adjusted based on the iterative density estimation, so that the effect of accurately predicting data is achieved.
In a preferred embodiment, the membership function similarity calculation in step S2 includes the following steps:
s2.1, fusing fuzzy rules with high similarity, and calculating the following formula in a fuzzy set similarity calculation mode based on a Gaussian membership function:
wherein cwin，j and ci，jRespectively indicating that the winning rule win and the ith rule are centered in the jth dimension,andrespectively representing the variances of the winning rule win and the ith rule in the jth dimension;
s2.2, taking the maximum membership degree of the intersection point coordinates of at least two membership degree functions as the similarity of the subsets, calculating the similarity of all the subsets of the front piece in the fuzzy rule according to the following formula, and aggregating to obtain the overlapping degree of the winning rule win and the rule i:
s2.3 to ensure the specificity of fuzzy rules, according to Si(win, i) > 0 and overlapwin，iSelecting a rule with rule similarity greater than a set value threrespap for fusion;
and S2.4, if the inequality condition of the step D is met, fusing the fuzzy rule by using a third rule algorithm.
In a preferred embodiment, the calculating step of the third rule algorithm for fusing fuzzy rules comprises:
s2.5 fuses the fuzzy rules according to the following formula:
respectively acquiring the fused rule centers CmergedVariance σ2 mergedAnd a back-part parameter omegamerged，
wherein Andrespectively representing the number of samples in the winning rule and the ith rule;andrespectively representing the variances of the winning rule and the ith rule; omegawin and ΩiThe parameters of the latter part of the winning rule win and the ith rule are respectively represented.
In another preferred embodiment, the fuzzy weight recursive least squares algorithm calculating step in step S3 includes:
s3.1 updating the back-piece parameters based on the following formula:
wherein ,i is an identity matrix of a back-part parameter, A represents a forgetting factor, omegaNRepresenting the nth iteration of the back-piece parameter. The fuzzy weight recursive least square algorithm is an extension of the recursive least square algorithm, and the method can keep the weight vector in a small bounded interval and realize higher generalization of the classifier.
In still another preferred or optional embodiment, in step S4, the step of constructing the prediction interval based on the boottrap method includes:
s4.1, constructing a prediction interval based on the following formula:
wherein , denotes theta independent repeatsPredicting the average value of output after the experiment; t is tθ/2(theta) is the (theta/2) quantile of the t-distribution function with theta degrees of freedom, sigma2 noisyRepresents the variance of the noise estimate in the data using the Gamma Test method.
Through the configuration of the preferred embodiment, compared with the prediction results of other interval prediction algorithms in the prior art, the results are as follows:
through the result comparison, the root mean square error, namely the standard error and the standard deviation of the result obtained by the load interval prediction method are obviously smaller than those of the three existing prediction algorithms, the average value is smaller, the result of the prediction interval is better, and the prediction time is also smaller than that of the existing algorithm by multiple. That is, the accuracy, stability and efficiency of the prediction result are obviously superior to those of the existing algorithm.
With the above configuration, the device load interval prediction method may be written in the form of a computer program and stored on a computer-readable storage medium, and when the computer program is executed by a processor, the device load interval prediction method as described above will be implemented.
Example one
The following describes, in detail, the accurate prediction of the load interval of the device based on the above configuration, with reference to the prediction of the load data carried by the air compressor industry as an example.
Taking an air compressor group system of a certain metallurgical enterprise as an example, load data of an air compressor group with 28 meshes from 2 months 1 days in 2018 to 2 months 2 in 2018 are selected, and the time interval is 1 min. The accuracy statistical index adopts root mean square error and is defined as follows:
the composite indicator CWC considers both coverage (PICP) and interval width (MPIW) for describing the reliability of the prediction interval, and is defined as follows:
CWC＝MPIW(1+γ(PICP)exp(-η(PICP-μ)))；
where η and μ are two hyperparameters γ (PICP) is defined as:
wherein ,Ui and LiRespectively, the upper and lower limits of the interval, when the target value falls within the constructed prediction interval, ci1, otherwise, ci＝0。CWCmedian and CWCsdThe mean and standard deviation were obtained by performing several experiments, respectively. CWC (continuous wave conductor)medianThe average value of the prediction intervals of the test structure performed a plurality of times is shown, and the smaller the value, the better the result of the prediction interval of the structure. CWC (continuous wave conductor)sdIndicates the stability of the structural prediction interval, and the smaller the value, the more stable the structural prediction interval performance.
It should be noted that the embodiments of the present invention have been described in terms of preferred embodiments, and not by way of limitation, and that those skilled in the art can make modifications and variations of the embodiments described above without departing from the spirit of the invention.
Claims (10)
1. A device load interval prediction method based on a fuzzy neural network comprises the following steps:
s1 initializing a rule base, calculating the potential of the rule based on an iterative density estimation method, and generating a fuzzy rule according to a first rule algorithm:
s2 calculating the similarity of the fuzzy rules based on the membership function similarity method, so as to fuse the fuzzy rules according to a third rule algorithm;
s3, updating the parameters of the fuzzy rule postware on line based on the fuzzy weight recursive least square algorithm;
s4 constructs a prediction interval based on the Bootstrap method.
2. The fuzzy neural network-based device load interval prediction method of claim 1, further comprising the steps of:
s1.1, based on a second rule algorithm, trimming and recalling fuzzy rules.
3. The fuzzy neural network-based device load interval prediction method of claim 2, wherein said second rule algorithm step comprises:
s1.1.1, calculating the closeness degree of the output data of the ith rule in the fuzzy rule and the real output data based on the following formula:
wherein d (N) represents the Nth real output data, yi(N) output data representing the ith rule; r isi back(N) indicating the proximity of the nth output data of the ith rule to the true output;
s1.1.2 calculating the Life Strength P of the ith rule based on the following formulai(N)：
Pi(N)＝ζ×Pi(N-1)+fi×ri back(N)
Wherein, the forgetting factor zeta belongs to (0, 1);
when the life intensity Pi(N) < ═ thresP, prune the ith rule from within the fuzzy rule, where thresP represents the rule pruning threshold:
s1.1.3 according to the formula:it is determined whether the ith rule is recalled, wherein,κ represents the number of pruned rules and the number of untrimmed rules in the rule base, respectively.
4. The fuzzy neural network-based device load interval prediction method of claim 1, wherein the potential step of calculating a rule based on the iterative density estimation method comprises:
the first-order Taylor expansion is carried out by utilizing the Gaussian membership function to obtain:
wherein Represents the nth sample;
the newly generated sample density is calculated based on the following formula:
wherein ：
5. the fuzzy neural network-based device load interval prediction method of claim 1, wherein said first rule algorithm step comprises:
s1.2 according to the conditions:orAt least one of, judging to generate fuzzy rules;
s1.3 calculation formula: cnew＝xN，ΩnewRand (u) to initialize the parameters of the newly generated fuzzy rule, where CnewDenotes the center, Ω, of the newly generated fuzzy rulenewA back-piece parameter representing the newly generated fuzzy rule.
6. The fuzzy neural network-based device load interval prediction method of claim 1, wherein the membership function similarity calculation comprises the steps of:
s2.1 calculation formula:
wherein cwin，j and ci，jRespectively indicating that the winning rule win and the ith fuzzy rule are centered in the jth dimension,andrespectively representing the variances of the winning rule win and the ith fuzzy rule in the jth dimension;
s2.2, taking the maximum membership degree of the intersection point coordinates of at least two membership degree functions as the similarity of the subsets, calculating the similarity of all the subsets of the front piece in the fuzzy rule according to the following formula, and aggregating to obtain the overlapping degree of the winning rule win and the rule i:
s2.3 according to Si(win, i) > 0 and overlapwin，iSelecting a rule with rule similarity greater than a set value threrespap;
and S2.4, if the inequality condition of the step S2.3 is met, fusing the fuzzy rule by using a third rule algorithm.
7. The fuzzy neural network-based device load interval prediction method of claim 1, wherein said third rule algorithm calculation step comprises:
s2.5 fuses the fuzzy rules according to the following formula:
respectively acquiring the Cfiltered and the variance sigma of the fused rule center2 mergedAnd a back-part parameter omegamerged，
wherein Andrespectively representing the number of samples in the winning rule and the ith fuzzy rule;andrespectively representing the variances of the winning rule and the ith fuzzy rule; omegawin and ΩiThe back-piece parameters of the winning rule win and the ith fuzzy rule are respectively represented.
8. The fuzzy neural network-based device load interval prediction method of claim 1, wherein the fuzzy weight recursive least square algorithm calculation step comprises:
s3.1 updating the back-piece parameters based on the following formula:
wherein ,i is an identity matrix of a back-part parameter, A represents a forgetting factor, omegaNRepresenting the nth iteration of the back-piece parameter.
9. The fuzzy neural network-based device load interval prediction method of claim 1, wherein the step of constructing the prediction interval based on the boottrap method comprises:
s4.1 constructs a prediction interval based on:
wherein , represents the average value of the predicted output after theta independent repeated experiments; t is tθ/2(theta) is the (theta/2) quantile of the t-distribution function with theta degrees of freedom, sigma2 noisyRepresents the variance of the noise estimate in the data using the Gamma Test method.
10. A computer-readable storage medium, on which a computer program is stored which, when being executed by a processor, carries out the method for device load interval prediction according to any one of claims 1 to 9.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910514109.XA CN110263410B (en) | 2019-06-14 | 2019-06-14 | Equipment load interval prediction method based on fuzzy neural network and computer readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910514109.XA CN110263410B (en) | 2019-06-14 | 2019-06-14 | Equipment load interval prediction method based on fuzzy neural network and computer readable storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110263410A true CN110263410A (en) | 2019-09-20 |
| CN110263410B CN110263410B (en) | 2023-04-28 |
Family
ID=67918313
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910514109.XA Active CN110263410B (en) | 2019-06-14 | 2019-06-14 | Equipment load interval prediction method based on fuzzy neural network and computer readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110263410B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113255238A (en) * | 2021-07-07 | 2021-08-13 | 南昌伐典热电科技有限责任公司 | Fuel gas utilization rate prediction method and system based on hybrid neural network |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109102066A (en) * | 2018-07-12 | 2018-12-28 | 山东师范大学 | Online time series data prediction technique, system and storage medium based on fuzzy reasoning |
| CN109697504A (en) * | 2018-11-09 | 2019-04-30 | 山东师范大学 | Time Series Forecasting Methods and device based on intuition circulation fuzzy neural network |
-
2019
- 2019-06-14 CN CN201910514109.XA patent/CN110263410B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109102066A (en) * | 2018-07-12 | 2018-12-28 | 山东师范大学 | Online time series data prediction technique, system and storage medium based on fuzzy reasoning |
| CN109697504A (en) * | 2018-11-09 | 2019-04-30 | 山东师范大学 | Time Series Forecasting Methods and device based on intuition circulation fuzzy neural network |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN113255238A (en) * | 2021-07-07 | 2021-08-13 | 南昌伐典热电科技有限责任公司 | Fuel gas utilization rate prediction method and system based on hybrid neural network |
| CN113255238B (en) * | 2021-07-07 | 2021-11-12 | 华能东莞燃机热电有限责任公司 | Fuel gas utilization rate prediction method and system based on hybrid neural network |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110263410B (en) | 2023-04-28 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| Almalaq et al. | Evolutionary deep learning-based energy consumption prediction for buildings | |
| Mason et al. | Predicting host CPU utilization in the cloud using evolutionary neural networks | |
| Wu et al. | Evolving RBF neural networks for rainfall prediction using hybrid particle swarm optimization and genetic algorithm | |
| Li et al. | Building's electricity consumption prediction using optimized artificial neural networks and principal component analysis | |
| Juang et al. | Hierarchical cluster-based multispecies particle-swarm optimization for fuzzy-system optimization | |
| Thong et al. | A new approach to multi-variable fuzzy forecasting using picture fuzzy clustering and picture fuzzy rule interpolation method | |
| Lu et al. | Generalized radial basis function neural network based on an improved dynamic particle swarm optimization and AdaBoost algorithm | |
| Deng et al. | Simulation-based evaluation of defuzzification-based approaches to fuzzy multiattribute decision making | |
| Choong et al. | State-of-the-art for modelling reservoir inflows and management optimization | |
| Orouskhani et al. | Evolutionary dynamic multi-objective optimization algorithm based on Borda count method | |
| Shen et al. | Multi-objective particle swarm optimization based on fuzzy optimality | |
| Li et al. | An optimisation method for complex product design | |
| Alhussan et al. | Wind speed forecasting using optimized bidirectional LSTM based on dipper throated and genetic optimization algorithms | |
| Ma | An Efficient Optimization Method for Extreme Learning Machine Using Artificial Bee Colony. | |
| Ghanbari et al. | An intelligent load forecasting expert system by integration of ant colony optimization, genetic algorithms and fuzzy logic | |
| Choong et al. | A dynamic fuzzy‐based dance mechanism for the bee colony optimization algorithm | |
| Bekker | Applying the cross-entropy method in multi-objective optimisation of dynamic stochastic systems | |
| CN110837857A (en) | Industrial electricity load prediction method, system and storage medium thereof | |
| Oubelaid et al. | Energy efficiency modeling using whale optimization algorithm and ensemble model | |
| Handoyo et al. | The fuzzy inference system with rule bases generated by using the fuzzy C-means to predict regional minimum wage in Indonesia | |
| Lian et al. | Training algorithm for radial basis function neural network based on quantum-behaved particle swarm optimization | |
| CN114240687A (en) | Energy hosting efficiency analysis method suitable for comprehensive energy system | |
| Worasucheep | Forecasting currency exchange rates with an Artificial Bee Colony-optimized neural network | |
| CN110263410A (en) | Machine utilization interval prediction method and computer readable storage medium based on fuzzy neural network | |
| CN111144611A (en) | Spatial load prediction method based on clustering and nonlinear autoregression |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |