CN116880201A - Water network channel state control system based on fuzzy neural network - Google Patents
Water network channel state control system based on fuzzy neural network Download PDFInfo
- Publication number
- CN116880201A CN116880201A CN202310945481.2A CN202310945481A CN116880201A CN 116880201 A CN116880201 A CN 116880201A CN 202310945481 A CN202310945481 A CN 202310945481A CN 116880201 A CN116880201 A CN 116880201A
- Authority
- CN
- China
- Prior art keywords
- water level
- fuzzy rule
- ith
- representing
- fuzzy
- 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
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 title claims abstract description 109
- 238000013528 artificial neural network Methods 0.000 title claims abstract description 21
- 230000000946 synaptic effect Effects 0.000 claims description 45
- 238000012545 processing Methods 0.000 claims description 41
- 210000002569 neuron Anatomy 0.000 claims description 35
- 238000000034 method Methods 0.000 claims description 18
- 239000011159 matrix material Substances 0.000 claims description 11
- 238000012549 training Methods 0.000 claims description 9
- 230000004913 activation Effects 0.000 claims description 6
- 238000005259 measurement Methods 0.000 claims description 6
- 210000004205 output neuron Anatomy 0.000 claims description 3
- 238000012937 correction Methods 0.000 claims description 2
- 230000006870 function Effects 0.000 abstract description 42
- 230000007246 mechanism Effects 0.000 abstract description 6
- 230000008569 process Effects 0.000 description 7
- 230000009286 beneficial effect Effects 0.000 description 5
- 238000011161 development Methods 0.000 description 4
- 238000004364 calculation method Methods 0.000 description 3
- 230000001419 dependent effect Effects 0.000 description 3
- 238000005516 engineering process Methods 0.000 description 3
- 238000010801 machine learning Methods 0.000 description 3
- 238000010248 power generation Methods 0.000 description 3
- 230000005540 biological transmission Effects 0.000 description 2
- 238000009826 distribution Methods 0.000 description 2
- 238000001914 filtration Methods 0.000 description 2
- 238000012417 linear regression Methods 0.000 description 2
- 238000007477 logistic regression Methods 0.000 description 2
- 230000009471 action Effects 0.000 description 1
- 238000013459 approach Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 238000013461 design Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 230000002349 favourable effect Effects 0.000 description 1
- 238000010348 incorporation Methods 0.000 description 1
- 230000002262 irrigation Effects 0.000 description 1
- 238000003973 irrigation Methods 0.000 description 1
- 238000004519 manufacturing process Methods 0.000 description 1
- 238000012544 monitoring process Methods 0.000 description 1
- 238000013486 operation strategy Methods 0.000 description 1
- 238000005457 optimization Methods 0.000 description 1
- 230000010355 oscillation Effects 0.000 description 1
- 230000003071 parasitic effect Effects 0.000 description 1
- 238000000611 regression analysis Methods 0.000 description 1
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
- G05B13/00—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion
- G05B13/02—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric
- G05B13/04—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators
- G05B13/042—Adaptive control systems, i.e. systems automatically adjusting themselves to have a performance which is optimum according to some preassigned criterion electric involving the use of models or simulators in which a parameter or coefficient is automatically adjusted to optimise the performance
Landscapes
- Engineering & Computer Science (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Automation & Control Theory (AREA)
- Feedback Control In General (AREA)
Abstract
The invention provides a water network channel state control system based on a fuzzy neural network, which belongs to the technical field of water network channel state control, and aims at the problem of low channel control precision at present, forms data-driven fuzzy control, obtains prediction model parameters through sample learning of mechanism driving control and through regression model and feedback adjustment, adopts a state space to combine with the fuzzy neural network to realize a multi-input multi-output state-control decision mechanism, adopts a regression model to the sample of the mechanism control, combines with the application of a hydrodynamic model to form a prediction function model, establishes a state space, adopts the fuzzy neural network for input, and realizes the prediction function fuzzy control based on the state space model.
Description
Technical Field
The invention belongs to the technical field of water network channel state control, and particularly relates to a water network channel state control system based on a fuzzy neural network.
Background
The water resource is closely related to the social development production, the shortage of the water resource and the uneven space-time distribution become important factors for restricting the social development, and the development of water conservancy automation can effectively improve the water resource transmission and distribution efficiency and reduce the water transmission loss of irrigation channels, thereby relieving the social contradiction caused by the shortage of the water resource. Since the 30 s of the 20 th century, the technology of hydraulic automatic gate has gradually developed, and the technology of channel automatic control has matured, and the core of channel automatic control is a control algorithm, namely a method for describing the logical relationship between channel input (water level or flow error) and output (gate action).
With the development of information technology, industries are rapidly developing towards digital, informationized, networked and intelligent industry modes, and machine learning has become a common and indispensable solution to data driving problems in most of science. However, these data-driven problems inherently present uncertainties in the data acquisition process. In particular, data collected from sensors in engineering is subject to various uncertainties due to measurement errors, knowledge imperfections, subject differences, and the like. Various forms of uncertainty may reduce the effectiveness and accuracy of intelligent decisions.
Fuzzy inference systems have proven to be a powerful and efficient method of dealing with uncertainty, using fuzzy logic to measure values with incomplete or uncertain information. Therefore, the Fuzzy Neural Network (FNN) combines the fuzzy reasoning system and the neural network, and can well solve the machine learning task in the uncertain big data environment. Therefore, how to solve the problem of low channel control accuracy by using machine learning is to be solved.
Disclosure of Invention
Aiming at the defects in the prior art, the water network channel state control system based on the fuzzy neural network solves the problem of low channel control precision at present.
In order to achieve the above purpose, the invention adopts the following technical scheme:
the utility model provides a water network channel state control system based on fuzzy neural network, include:
the first processing module is used for acquiring a water level data sample through a sensor operated by the water network;
the second processing module is used for correcting deviation and error of the sample by utilizing a tracking differentiator of the fhan function to obtain detected water level data;
the third processing module is used for processing the water level data by using the FCM clustering method;
the fourth processing module is used for training the synaptic weight by utilizing a RENNCOM algorithm according to the water level data after clustering;
the fifth processing module is used for setting an objective function according to the synaptic weight obtained through training, setting a state equation of the synaptic weight and a result parameter of an output equation to obtain a prediction model, wherein the result parameter comprises a parameter vector of each fuzzy rule;
and the sixth processing module is used for obtaining a predicted water level by utilizing a prediction model according to the water level at the current moment, inputting the predicted water level to the gate opening controller, carrying out gate opening scheduling, and forming the state control of the water network channel according to the relation among the water level, the flow and the gate opening control and the variable influence under different working conditions.
The beneficial effects of the invention are as follows: aiming at the problem of low control precision of the current channel, data-driven fuzzy control is formed, the prediction model parameters are obtained through sample learning of mechanism-driven control and regression model and feedback adjustment, a state space is combined with a fuzzy neural network to realize a multi-input multi-output state-control decision mechanism, a state space is established, and the fuzzy neural network is adopted for input to realize the fuzzy control of the prediction function based on the state space model. The invention solves the problem of low control precision of the current channel.
Further, in the second processing module, the deviation and the error of the sample are corrected by using a tracking differentiator of the fhan function, and the expression is as follows:
wherein , and />Respectively represent x 1 and x2 X is the first derivative of 1 and x2 Representing the tracking signal and the differential signal, respectively, fhan (·) representing the fhan function, v representing the input signal, r 0 Represents a velocity factor, h 0 Representing the filter factor;
the discrete form is as follows:
wherein ,x1 (k ') represents the tracking signal at time k', v (k ') represents the input signal at time k', x 2 (k ') represents the differential signal at time k', x 1 (k '+1) represents a tracking signal at time k' +1, h represents a filter factor, and x 2 (k '+1) represents a differential signal at time k' +1;
its fhan (x) 1 (k')-v(k'),x 2 (k'),r 0 ,h 0 ) The specific form of (2) is as follows:
wherein ,d、a0 、a 1 、a 2 、a、y、s y and sa All represent tracking differentiator adjustable parameters.
The beneficial effects of the above-mentioned further scheme are: according to the invention, through carrying out deviation and error correction processing on the sample, the interference on water level data in the data acquisition process is reduced, and the accurate water level data result is obtained.
Still further, the water level data is processed by using the FCM clustering method, and the expression is as follows:
wherein ,σij Represents standard deviation of membership function, P represents total number of water level data sample set, u in Represents membership, r represents the total number of clusters, m i'j Column data representing row i and column j in m-dimensional water level sample dataset, x j' (n) represents the j' th input vector, m kj Represents the kth row and jth column data in an m-dimensional water level sample dataset, u in (n) represents the membership degree of the nth water level data sample belonging to the ith cluster, and c represents the number of ambiguity parameters.
The beneficial effects of the above-mentioned further scheme are: the invention processes the water level data by using the FCM clustering method, has high convergence rate, can quickly cluster large-scale data, and has low requirement on equipment computing power.
Still further, the fourth processing module includes:
a first processing unit for determining synaptic weight BDRNN outputs for N neurons by:
wherein , and />Respectively represent the kth-1 and kth neuron outputs formed when the nth water level data sample is processed under the ith' fuzzy rule, f 1 Represents the activation function, j "represents the dimension index of the input vector, m represents the dimension of the water level sample dataset,/v> and />Respectively represent the synaptic weights, x, of the k-1 th and k-th neurons under the ith fuzzy rule j' (n) represents the j' th input vector, "> and />All represent the feedback weights of the kth neuron of the synaptic weight BDRNN under the ith fuzzy rule, ++>Represents the (k-1) th neuron formed when the (n-1) th water level data sample is processed under the (i) < th > fuzzy rule, <>The (1) th neuron formed when the (1) th water level data sample is processed under the (i) th fuzzy rule is represented, r represents the total number of clusters, the (i) th fuzzy rule corresponds to the (i) th cluster, and N represents the (N) th neuron;
the second processing unit is used for calculating and obtaining the output of the synaptic weight BDRNN according to the determined output of the synaptic weight BDRNN for the ith fuzzy rule:
wherein ,gi” (n) represents the synaptic weight BDRNN output under the ith fuzzy rule, f 2 Representing an activation function, b ij Representing the synaptic weights of the output neurons,a neuron output representing a kth block formed when the nth water level data sample is processed under the ith "fuzzy rule;
the third processing unit is configured to complete training of the synaptic weights by forming the feedback weights of each block according to the output of the synaptic weights BDRNN obtained by the fifth processing unit by using the following feedback matrix:
wherein ,representing the feedback matrix under the ith "fuzzy rule.
The beneficial effects of the above-mentioned further scheme are: the present invention may incorporate various constraints on the learning process by training synaptic weights. In this case, the gradient descent based approach cannot guarantee stable learning because there is a feedback connection at the hidden layer of the BDRNN. Thus, constraints related to stable learning can be introduced in a proper functional form and optimized simultaneously with the standard error function.
Still further, the constraint condition of the eigenvalues of the feedback matrix is as follows:
wherein ,λi”k Representing under the ith "fuzzy ruleCharacteristic value of> and />All represent the feedback weights of the kth neuron of the synaptic weight BDRNN under the ith fuzzy rule.
Still further, the expression of the objective function of the predictive model is as follows:
wherein ,pik and fd (z ik ) All represent the objective function of the predictive model, e represents the natural logarithm, a s Represents the slope, lambda of the sigmoid function i”k Representing under the ith "fuzzy ruleCharacteristic value of> and />All represent the feedback weights of the kth neuron of the synaptic weight BDRNN under the ith fuzzy rule.
Still further, the constraint condition of the objective function is:
(1) Setting error minimization, taking mean square error MSE as a measurement:
wherein P represents the total number of water level data sample sets, n' represents time, y (n) andtime series respectively representing water level and flow rate;
(2) Setting a compensation function phi to be minimized:
wherein r represents the total number of clusters, i represents the number of clusters, N represents the total number of water level time series data, k represents the number of neurons, and p ik The stability function is represented by a function of the stability, and />Feedback weights of the kth neuron of the synaptic weight BDRNN under the ith fuzzy rule are respectively represented;
(3) Setting a weight space search condition phi w :
Φ w =dθ T ·(Δ 2 ) -1 ·dθ-1=0
Where dθ represents the amount of variation of the resulting parameter vector at each iteration, Δ represents the diagonal matrix containing the maximum parameter variation for each weight, and T represents the transpose.
The beneficial effects of the above-mentioned further scheme are: according to the invention, through the three set targets, the finally obtained prediction model has the characteristic of minimum error.
Drawings
Fig. 1 is a schematic diagram of a system structure according to the present invention.
Detailed Description
The following description of the embodiments of the present invention is provided to facilitate understanding of the present invention by those skilled in the art, but it should be understood that the present invention is not limited to the scope of the embodiments, and all the inventions which make use of the inventive concept are protected by the spirit and scope of the present invention as defined and defined in the appended claims to those skilled in the art.
Examples
As shown in fig. 1, the present invention provides a water network channel state control system based on a fuzzy neural network, comprising:
a first processing module for acquiring water level data samples through sensors operated by a water network
In the embodiment, sensor data of water network operation is selected as a sample, water level values of a plurality of points in each sample at the same time are taken as independent variables, and gate control guide corresponding to each sample is taken as the dependent variable; the argument is an input value of the system, i.e., a value input to the first module; the dependent variable is the final output value of the system, namely gate opening control of the output after the sixth module processing is completed.
In this embodiment, sensor data of water network operation is selected as a sample, supervised learning is performed, on one hand, the problem of prediction is solved from a regression model, a water level value of a plurality of points in a control process in each sample at the same time is taken as an independent variable, a gate control guide corresponding to the water level value is taken as a dependent variable, and an applicable regression analysis method includes, but is not limited to: linear Regression linear Regression, logistic Regression logistic Regression, polynomial Regression polynomial Regression, stepwise Regression stepwise Regression, ridge Regression, lasso Regression, elastic Net Regression, and the like.
The second processing module is used for correcting deviation and error of the sample by utilizing a tracking differentiator of the fhan function to obtain detected water level data;
in the embodiment, the invention introduces a tracking differentiator based on the fhan function, solves the problems of errors, noise and the like by using the tracking and filtering functions of the tracking differentiator, and further rapidly tracks an original signal by using the output signal without overshoot to form closed-loop control of an observer and form a sample favorable for learning. The calculation adopts the following algorithm:
the discrete form is as follows:
wherein , and />Respectively represent x 1 and x2 X is the first derivative of 1 and x2 Representing the tracking signal and the differential signal, respectively, fhan (·) representing the fhan function, v representing the input signal, r 0 Represents a velocity factor, h 0 Representing the filter factor, fh representing the fhan function, x 1 (k ') represents the tracking signal at time k', v (k ') represents the input signal at time k', x 2 (k ') represents the differential signal at time k', x 1 (k '+1) represents a tracking signal at time k' +1, h represents a filter factor, and x 2 (k '+1) represents a differential signal at time k' +1.
Its fhan (x) 1 (k')-v(k'),x 2 (k'),r 0 ,h 0 ) The specific form is as follows:
wherein ,d、a0 、a 1 、a 2 、a、y、s y and sa All represent tracking differentiator adjustable parameters;
reasonable parameters are selected, a tracking differentiator is accurately utilized, and deviation and error generated in the linkage control process of mechanism driving are corrected to obtain the final detected water level.
The third processing module is used for processing the water level data by using the FCM clustering method;
in this embodiment, the FCM clustering method is used to process the water level data of the data samples, and the most suitable cluster center is provided according to the minimum distance between the data samples belonging to each cluster, where each cluster center is concentrated on a part of the data set, and for a given number of clusters and data sets of samples, the cluster center is represented by the following formula:
and carrying out standard deviation calculation of the membership function to determine that the precondition part of the fuzzy rule remains unchanged, wherein the calculation formula is as follows:
wherein ,σij Represents standard deviation of membership function, P represents total number of water level data sample set, u in Represents membership, r represents the total number of clusters, m i'j Column data representing row i and column j in m-dimensional water level sample dataset, x j' (n) represents the j' th input vector, m kj Represents the kth row and jth column data in an m-dimensional water level sample dataset, u in (n) represents the membership degree of the nth water level data sample belonging to the ith cluster, and c represents the ambiguity parameter.
The fourth processing module is configured to train the synaptic weights by using a RENNCOM algorithm according to the clustered water level data, and includes:
a first processing unit for determining synaptic weight BDRNN outputs for N neurons by:
wherein , and />Respectively represent the kth-1 and kth neuron outputs formed when the nth water level data sample is processed under the ith' fuzzy rule, f 1 Represents the activation function, j "represents the dimension index of the input vector, m represents the dimension of the water level sample dataset,/v> and />Respectively represent the synaptic weights, x, of the k-1 th and k-th neurons under the ith fuzzy rule j' (n) represents the j' th input vector, "> and />All represent the feedback weights of the kth neuron of the synaptic weight BDRNN under the ith fuzzy rule, ++>Represents the (k-1) th neuron formed when the (n-1) th water level data sample is processed under the (i) < th > fuzzy rule, <>The (k) th neuron formed when the (n-1) th water level data sample is processed under the (i) th fuzzy rule, and r is the total number of clusters, wherein the (i) th fuzzy ruleN represents the N-th neuron corresponding to the i-th cluster;
the second processing unit is used for calculating and obtaining the output of the synaptic weight BDRNN according to the determined output of the synaptic weight BDRNN for the ith fuzzy rule:
wherein ,gi” (n) represents the synaptic weight BDRNN output under the ith fuzzy rule, f 2 Representing an activation function, b ij Representing the synaptic weights of the output neurons,a neuron output representing a kth block formed when the nth water level data sample is processed under the ith "fuzzy rule;
the third processing unit is configured to complete training of the synaptic weights by forming the feedback weights of each block according to the output of the synaptic weights BDRNN obtained by the fifth processing unit by using the following feedback matrix:
wherein ,representing the feedback matrix under the ith "fuzzy rule.
The fifth processing module is used for setting an objective function according to the synaptic weight obtained through training, and setting a state equation of the synaptic weight and a result parameter of an output equation, wherein the result parameter comprises a parameter vector of each fuzzy rule;
in this embodiment, an objective function is set. The objective function that is suitable for incorporation into the constraint equation is a sigmoid function that is smooth and continuously differentiable and therefore robust to problems caused by parasitic oscillations. Thus, the expression of the objective function of the predictive model is as follows:
wherein ,pik and fd (z ik ) All represent the objective function of the predictive model, e represents the natural logarithm, a s Represents the slope, lambda of the sigmoid function i”k Representing under the ith "fuzzy ruleCharacteristic value of> and />All represent the feedback weights of the kth neuron of the synaptic weight BDRNN under the ith fuzzy rule. When the characteristic value is in the unit circle, the model is stable, so that the value of the active area as is [4,8 ]]Within the range.
In the present embodiment, the state equation and output equation result parameters of BDRNN are set to include the parameter vector θ of each fuzzy rule con . The following three targets of the objective function are realized by utilizing RENNCOM algorithm:
(1) Setting error minimization, taking mean square error MSE as a measurement:
wherein P represents the total number of water level data sample sets, n' represents time, y (n) andtime series respectively representing water level and flow rate;
(2) Setting a compensation function phi to be minimized:
wherein r represents the total number of clusters, i represents the number of clusters, N represents the total number of water level time series data, k represents the number of neurons, and p ik The stability function is represented by a function of the stability, and />Feedback weights of the kth neuron of the synaptic weight BDRNN under the ith fuzzy rule are respectively represented;
(3) Setting a weight space search condition phi w :
Φ w =dθ T ·(Δ 2 ) -1 ·dθ-1=0
Where dθ represents the amount of variation of the resulting parameter vector at each iteration, Δ represents the diagonal matrix containing the maximum parameter variation for each weight, and T represents the transpose.
And the sixth processing module is used for inputting the output predicted water level to the gate opening controller according to the processing result of the fifth processing module, carrying out gate opening scheduling, and forming the state control of the water network channel according to the relation among the water level, the flow and the gate opening control and the variable influence under different conditions.
In this embodiment, the output predicted water level is input to the gate opening controller to perform gate opening scheduling, and according to the relationship among the water level, the flow rate, and the gate opening control, and the variable influence under different working conditions, the operation is finally realized by checking, feeding back, adjusting, perfecting.
In the embodiment, the fuzzy control based on the rule is finally formed through the design, the rule between the input and the output is formulated, the output value is obtained according to the rule judgment of the input, the input water level value is realized, the control guidance is output, finally, the prediction function controller based on the state space model is designed based on the prediction function control algorithm principle, and the problem that the previous learning cannot be applied once the state is changed, such as the flow, the water level and the water demand target is solved. Accurate control operation output obtained under different working conditions is realized.
In this embodiment, the present invention proposes an application innovation in the fuzzy control in the water supply-power generation channel. And a multi-input multi-output state space for the linkage control of the water supply and power generation channels is formed, the water levels of all measuring points at the same moment are used as the opening control operation guide of all gates obtained by multi-input and are used as multi-output, and the efficient and stable connection of the control between different working conditions is ensured. The deep learning is innovated to achieve an optimization method and an operation strategy, and a fuzzy neural network is introduced to perform hierarchical classification on the control object, so that correct and reasonable decision guidance is further obtained from a state space, and a prediction function controller based on a state space model is designed based on the principle of a prediction function control algorithm.
In the embodiment, the invention provides an anti-disturbance momentum measurement innovation in water supply-power generation channel control monitoring. And the initial data of the channel measurement device is subjected to an anti-interference and filtering algorithm, so that errors are avoided, and feedback is true and accurate. The accuracy is further improved, the accurate real-time water level and flow can be fed back to the regulator, and the linkage control is supported.
Claims (7)
1. A water network channel state control system based on a fuzzy neural network, comprising:
the first processing module is used for acquiring a water level data sample through a sensor operated by the water network;
the second processing module is used for correcting deviation and error of the sample by utilizing a tracking differentiator of the fhan function to obtain detected water level data;
the third processing module is used for processing the water level data by using the FCM clustering method;
the fourth processing module is used for training the synaptic weight by utilizing a RENNCOM algorithm according to the water level data after clustering;
the fifth processing module is used for setting an objective function according to the synaptic weight obtained through training, setting a state equation of the synaptic weight and a result parameter of an output equation to obtain a prediction model, wherein the result parameter comprises a parameter vector of each fuzzy rule;
and the sixth processing module is used for obtaining a predicted water level by utilizing a prediction model according to the water level at the current moment, inputting the predicted water level to the gate opening controller, carrying out gate opening scheduling, and forming the state control of the water network channel according to the relation among the water level, the flow and the gate opening control and the variable influence under different working conditions.
2. The water network channel state control system based on a fuzzy neural network of claim 1, wherein in the second processing module, a tracking differentiator of fhan function is used to perform deviation and error correction on the sample, and the expression is as follows:
wherein , and />Respectively represent x 1 and x2 X is the first derivative of 1 and x2 Representing the tracking signal and the differential signal, respectively, fhan (·) representing the fhan function, v representing the input signal, r 0 Represents a velocity factor, h 0 Representing the filter factor;
the discrete form is as follows:
wherein ,x1 (k ') represents the tracking signal at time k', v (k ') represents the input signal at time k', x 2 (k ') represents the differential signal at time k', x 1 (k '+1) represents a tracking signal at time k' +1, h represents a filter factor, and x 2 (k' +1) tableA differential signal at time k' +1;
its fhan (x) 1 (k')-v(k'),x 2 (k'),r 0 ,h 0 ) The specific form of (2) is as follows:
wherein ,d、a0 、a 1 、a 2 、a、y、s y and sa All represent tracking differentiator adjustable parameters.
3. The water network channel state control system based on a fuzzy neural network according to claim 1, wherein the water level data is processed by using an FCM clustering method, and the expression is as follows:
wherein ,σij Represents standard deviation of membership function, P represents total number of water level data sample set, u in Represents membership, r represents the total number of clusters, m i'j Column data representing row i and column j in m-dimensional water level sample dataset, x j' (n) represents the j' th input vector, m kj Represents the kth row and jth column data in an m-dimensional water level sample dataset, u in (n) represents the membership degree of the nth water level data sample belonging to the ith cluster, and c represents the ambiguity parameter.
4. The water network channel state control system based on a fuzzy neural network of claim 1, wherein the fourth processing module includes:
a first processing unit for determining synaptic weight BDRNN outputs for N neurons by:
wherein , and />Respectively represent the kth-1 and kth neuron outputs formed when the nth water level data sample is processed under the ith' fuzzy rule, f 1 Represents the activation function, j "represents the dimension index of the input vector, m represents the dimension of the water level sample dataset,/v> and />Respectively represent the synaptic weights, x, of the k-1 th and k-th neurons under the ith fuzzy rule j' (n) represents the j' th input vector, "> and />All represent the feedback weights of the kth neuron of the synaptic weight BDRNN under the ith fuzzy rule, ++>Represents the (k-1) th neuron formed when the (n-1) th water level data sample is processed under the (i) < th > fuzzy rule, <>The (1) th neuron formed when the (1) th water level data sample is processed under the (i) th fuzzy rule is represented, r represents the total number of clusters, the (i) th fuzzy rule corresponds to the (i) th cluster, and N represents the (N) th neuron;
the second processing unit is used for calculating and obtaining the output of the synaptic weight BDRNN according to the determined output of the synaptic weight BDRNN for the ith fuzzy rule:
wherein ,gi” (n) represents the synaptic weight BDRNN output under the ith fuzzy rule, f 2 Representing an activation function, b ij Representing the synaptic weights of the output neurons,a neuron output representing a kth block formed when the nth water level data sample is processed under the ith "fuzzy rule;
the third processing unit is configured to complete training of the synaptic weights by forming the feedback weights of each block according to the output of the synaptic weights BDRNN obtained by the fifth processing unit by using the following feedback matrix:
wherein ,represent the firsti' feedback matrix under fuzzy rule.
5. The water network channel state control system based on a fuzzy neural network of claim 4, wherein the constraints of the eigenvalues of the feedback matrix are as follows:
wherein ,λi”k Representing under the ith "fuzzy ruleCharacteristic value of> and />All represent the feedback weights of the kth neuron of the synaptic weight BDRNN under the ith fuzzy rule.
6. The water network channel state control system based on a fuzzy neural network of claim 5, wherein the expression of the objective function of the predictive model is as follows:
wherein ,pik and fd (z ik ) All represent the objective function of the predictive model, e represents the natural logarithm, a s Represents the slope, lambda of the sigmoid function i”k Representing under the ith "fuzzy ruleCharacteristic value of> and />All represent the feedback weights of the kth neuron of the synaptic weight BDRNN under the ith fuzzy rule.
7. The water network channel state control system based on a fuzzy neural network of claim 6, wherein the constraint of the objective function is:
(1) Setting error minimization, taking mean square error MSE as a measurement:
wherein P represents the total number of water level data sample sets, n' represents time, y (n) andtime series respectively representing water level and flow rate;
(2) Setting a compensation function phi to be minimized:
wherein r represents the total number of clusters, i represents the number of clusters, N represents the total number of water level time series data, k represents the number of neurons, and p ik The stability function is represented by a function of the stability, and />Feedback weights of the kth neuron of the synaptic weight BDRNN under the ith fuzzy rule are respectively represented;
(3) Setting a weight space search condition phi w :
Φ w =dθ T ·(Δ 2 ) -1 ·dθ-1=0
Where dθ represents the amount of variation of the resulting parameter vector at each iteration, Δ represents the diagonal matrix containing the maximum parameter variation for each weight, and T represents the transpose.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310945481.2A CN116880201A (en) | 2023-07-28 | 2023-07-28 | Water network channel state control system based on fuzzy neural network |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310945481.2A CN116880201A (en) | 2023-07-28 | 2023-07-28 | Water network channel state control system based on fuzzy neural network |
Publications (1)
Publication Number | Publication Date |
---|---|
CN116880201A true CN116880201A (en) | 2023-10-13 |
Family
ID=88271379
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310945481.2A Pending CN116880201A (en) | 2023-07-28 | 2023-07-28 | Water network channel state control system based on fuzzy neural network |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116880201A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117824788A (en) * | 2024-03-05 | 2024-04-05 | 河海大学 | Water level monitoring and analyzing system |
-
2023
- 2023-07-28 CN CN202310945481.2A patent/CN116880201A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117824788A (en) * | 2024-03-05 | 2024-04-05 | 河海大学 | Water level monitoring and analyzing system |
CN117824788B (en) * | 2024-03-05 | 2024-05-28 | 河海大学 | Water level monitoring and analyzing system |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN109685252B (en) | Building energy consumption prediction method based on cyclic neural network and multi-task learning model | |
CN101630376B (en) | Soft-sensing modeling method and soft meter of multi-model neural network in biological fermentation process | |
CN112884056A (en) | Optimized LSTM neural network-based sewage quality prediction method | |
CN109472057B (en) | Product processing quality prediction device and method based on cross-process implicit parameter memory | |
CN110619360A (en) | Ultra-short-term wind power prediction method considering historical sample similarity | |
CN111650834B (en) | Sewage treatment process prediction control method based on extreme learning machine | |
CN111144552B (en) | Multi-index grain quality prediction method and device | |
CN116880201A (en) | Water network channel state control system based on fuzzy neural network | |
Han et al. | Hierarchical nonlinear model predictive control with multi-time-scale for wastewater treatment process | |
CN110245398B (en) | Soft measurement deep learning method for thermal deformation of air preheater rotor | |
CN111290282A (en) | Predictive control method for thermal power generating unit coordination system | |
CN116865343B (en) | Model-free self-adaptive control method, device and medium for distributed photovoltaic power distribution network | |
CN112990601B (en) | Worm wheel machining precision self-healing system and method based on data mining | |
CN115206444A (en) | Optimal drug dosage prediction method based on FCM-ANFIS model | |
Yang et al. | Application of fuzzy neural network PID algorithm in oil pump control | |
CN117390960A (en) | Intelligent analysis method for machining precision of numerical control machine tool | |
Juan et al. | Multi-step prediction of dissolved oxygen in river based on random forest missing value imputation and attention mechanism coupled with recurrent neural network | |
CN113065281B (en) | TE process time sequence prediction method based on transfer entropy and long-short-term memory network | |
CN114943277A (en) | Method for identifying operation condition and predicting operation performance of comprehensive energy system | |
Prasad et al. | A Comprehensive overview on performance of cascaded three tank level system using neural network predictive controller | |
CN113935600A (en) | Deep learning-based adaptive economic dispatching system and method | |
CN111626465A (en) | New energy power short-term interval prediction method and system | |
Di et al. | Prediction of dissolved oxygen concentration based on improved RBF | |
Li et al. | Monitoring Model of Coal Mill in Power Plant Based on Big Data and BP Neural Network | |
CN115616978B (en) | SQ-LSTMA-based thermal error prediction model, prediction method and control system |
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 |