CN116070669B - Workshop energy consumption prediction method and management system based on improved deep belief network - Google Patents
Workshop energy consumption prediction method and management system based on improved deep belief network Download PDFInfo
- Publication number
- CN116070669B CN116070669B CN202310001073.1A CN202310001073A CN116070669B CN 116070669 B CN116070669 B CN 116070669B CN 202310001073 A CN202310001073 A CN 202310001073A CN 116070669 B CN116070669 B CN 116070669B
- Authority
- CN
- China
- Prior art keywords
- energy consumption
- consumption prediction
- data
- workshop
- network
- 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.)
- Active
Links
- 238000005265 energy consumption Methods 0.000 title claims abstract description 133
- 238000000034 method Methods 0.000 title claims abstract description 34
- 241000272201 Columbiformes Species 0.000 claims abstract description 30
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 25
- 238000013528 artificial neural network Methods 0.000 claims abstract description 21
- 238000005457 optimization Methods 0.000 claims abstract description 18
- 238000000354 decomposition reaction Methods 0.000 claims description 39
- 238000012544 monitoring process Methods 0.000 claims description 17
- 230000002159 abnormal effect Effects 0.000 claims description 7
- 238000012549 training Methods 0.000 claims description 5
- 238000010606 normalization Methods 0.000 claims description 3
- 238000000605 extraction Methods 0.000 claims description 2
- 238000013480 data collection Methods 0.000 claims 1
- 238000004364 calculation method Methods 0.000 description 7
- 238000012360 testing method Methods 0.000 description 6
- 230000008569 process Effects 0.000 description 5
- 238000012545 processing Methods 0.000 description 5
- 238000004590 computer program Methods 0.000 description 4
- 244000144992 flock Species 0.000 description 4
- 230000005856 abnormality Effects 0.000 description 3
- 238000012417 linear regression Methods 0.000 description 3
- 230000009467 reduction Effects 0.000 description 3
- 238000012706 support-vector machine Methods 0.000 description 3
- 238000010586 diagram Methods 0.000 description 2
- 238000004134 energy conservation Methods 0.000 description 2
- 230000006870 function Effects 0.000 description 2
- 238000007781 pre-processing Methods 0.000 description 2
- 230000011218 segmentation Effects 0.000 description 2
- 238000012935 Averaging Methods 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 238000000137 annealing Methods 0.000 description 1
- 230000005540 biological transmission Effects 0.000 description 1
- 230000008859 change Effects 0.000 description 1
- 238000004891 communication Methods 0.000 description 1
- 238000013527 convolutional neural network Methods 0.000 description 1
- 238000001816 cooling Methods 0.000 description 1
- 230000010485 coping Effects 0.000 description 1
- 238000003066 decision tree Methods 0.000 description 1
- 238000013135 deep learning Methods 0.000 description 1
- 230000007547 defect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 238000010801 machine learning Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000003672 processing method Methods 0.000 description 1
- 238000007670 refining Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/004—Artificial life, i.e. computing arrangements simulating life
- G06N3/006—Artificial life, i.e. computing arrangements simulating life based on simulated virtual individual or collective life forms, e.g. social simulations or particle swarm optimisation [PSO]
-
- 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
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q10/00—Administration; Management
- G06Q10/04—Forecasting or optimisation specially adapted for administrative or management purposes, e.g. linear programming or "cutting stock problem"
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06Q—INFORMATION AND COMMUNICATION TECHNOLOGY [ICT] SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES; SYSTEMS OR METHODS SPECIALLY ADAPTED FOR ADMINISTRATIVE, COMMERCIAL, FINANCIAL, MANAGERIAL OR SUPERVISORY PURPOSES, NOT OTHERWISE PROVIDED FOR
- G06Q50/00—Systems or methods specially adapted for specific business sectors, e.g. utilities or tourism
- G06Q50/04—Manufacturing
Abstract
The invention discloses a workshop energy consumption prediction method based on an improved deep confidence network, which comprises the following steps of: collecting energy consumption data of workshop equipment to form a data set; decomposing the data of the data sets by adopting an empirical mode and reconstructing the data based on sample entropy, so as to reduce randomness and volatility among different data sets; establishing a deep confidence network energy consumption prediction model optimized by a pigeon optimization algorithm, wherein the deep confidence network energy consumption prediction model adopts parameter selection of the pigeon optimization algorithm optimized deep confidence network to train the deep confidence network energy consumption prediction model; the trained deep confidence network energy consumption prediction model predicts workshop energy consumption, wherein the deep confidence network energy consumption prediction model consists of a plurality of limited Boltzmann machine stacks and BP neural networks. The invention also discloses a workshop energy consumption management system based on the improved deep belief network. The method accelerates the convergence speed of the algorithm and improves the accuracy of workshop energy consumption prediction.
Description
Technical Field
The invention relates to an energy consumption prediction method and an energy consumption management system, in particular to a workshop energy consumption prediction method and a workshop energy consumption management system based on an improved deep confidence network.
Background
The statistics and management of the energy consumption of the workshop are the basis for realizing the energy conservation and emission reduction of the workshop, but most of the statistics of the energy consumption of the workshop at present are realized by installing meters of different energy sources, then the statistics of the energy consumption of the workshop is realized manually, and then the energy consumption data of the workshop is analyzed, so that the purpose of reducing the consumption of the workshop is realized. However, the above method has the following disadvantages: (1) The manual meter reading statistics have certain errors, if the workshop is very large, a large amount of manpower is required for meter reading, and the manpower is wasted; (2) real-time transmission of workshop energy consumption data cannot be realized; (3) It is difficult to analyze the workshop energy consumption data finely, and the requirements of energy conservation and emission reduction of the workshop cannot be met.
The prediction of the energy consumption of the workshop is taken as one of important components of the consumption reduction of the workshop, and the current prediction method of the energy consumption of the workshop mainly comprises two types: the traditional machine learning method comprises multiple linear regression, decision tree, support vector machine and other methods, and the deep learning algorithm comprises an artificial neural network, a self-encoder, a deep confidence network, a convolutional neural network and the like.
The accuracy of predicting the workshop energy consumption by adopting the multiple linear regression algorithm is good, but the operation time of the model is long due to the fact that the related variables of the workshop energy consumption are numerous, meanwhile, the nonlinear problem cannot be solved by adopting the multiple linear regression algorithm, and a large dependency relationship exists between an input value and an output value of the model; the energy consumption of a workshop is predicted by adopting a support vector machine, and meanwhile, the parameters of the support vector machine are optimized by adopting an intelligent algorithm, but the algorithm has poor effect when dealing with large-scale calculated amount, and the model is complex and difficult to realize; the energy consumption of the workshop is predicted by adopting the artificial neural network, so that the method is suitable for processing nonlinear data, meanwhile, the artificial neural network has strong nonlinear fitting capacity and approximation capacity, but the processing correlation of the artificial neural network to the sequence data is insufficient, and the workshop energy consumption data with a certain time correlation cannot be processed.
Disclosure of Invention
Aiming at the defects in the prior art, the invention aims to provide a workshop energy consumption prediction method based on an improved deep confidence network, and the energy cost of enterprises is reduced by refining the call of the workshop energy consumption through energy consumption prediction. The invention further aims to provide a workshop energy consumption management system based on the improved deep confidence network, which is used for realizing real-time acquisition of workshop energy consumption data, finding out abnormal situations of workshop energy consumption in time, processing the abnormal situations and effectively reducing the workshop energy consumption.
The technical scheme of the invention is as follows: a workshop energy consumption prediction method based on an improved deep belief network comprises the following steps:
step 1, collecting energy consumption data of workshop equipment to form a data set;
step 2, decomposing the data of the data sets by adopting an empirical mode and reconstructing based on sample entropy, so as to reduce randomness and fluctuation among different data sets;
step 3, establishing a deep confidence network energy consumption prediction model optimized by a pigeon cluster optimization algorithm, optimizing parameter selection of the deep confidence network by the deep confidence network energy consumption prediction model by adopting the pigeon cluster optimization algorithm, and training the deep confidence network energy consumption prediction model by the data set processed in the step 2;
step 4, predicting workshop energy consumption by the trained deep confidence network energy consumption prediction model;
the deep confidence network energy consumption prediction model comprises a plurality of limited Boltzmann machine stacks and a BP neural network, wherein the plurality of limited Boltzmann machine stacks automatically extract characteristic values layer by layer of an input sequence through unsupervised learning, the extracted characteristic values are used as the BP neural network input into a top layer, the BP neural network trains and fits the characteristic values and output power of data to complete regression prediction, and meanwhile, the BP neural network adjusts all network parameters of the limited Boltzmann machines at the bottom layer.
Further, the data set is normalized prior to employing empirical mode decomposition and reconstruction based on sample entropy, the normalization normalizing the data to [0,1] as follows:
wherein X is max Represents the maximum value in the dataset, X min Representing minimum value in data set, X nor Representing the normalized value.
Further, empirical mode decomposition solves the modal aliasing phenomenon existing in modal decomposition by adding white noise to the original signal, however, the modal aliasing may not be completely solved when the amplitude of the added white noise is too low, too high amplitude of the added white noise may cause too many spurious components to appear in the original signal, and furthermore, each time white noise is added, one decomposition may be caused, and if the number of times of adding white noise is too large, the calculation amount of the decomposition may be increased. In the step 2, the step of decomposing the data of the data set by adopting an empirical mode comprises the following steps:
step 201, adding a white noise signal to an original signal to be decomposed;
step 202, adding two groups of positive and negative white noise signals w with equal absolute values to the original signal +(t) And w -(t) Then decomposing according to the following formula
Wherein d i，j(t) Representing the ith IMF component, p, obtained by the jth decomposition j(t) Representing the residual component obtained by the jth decomposition;representing the ith IMF component resulting from the decomposition, < >>Representing the residual component resulting from the decomposition;
step 203, calculating the average value of the two groups of IMF components obtained in the step 202;
step 204, performing EMD decomposition on the average value obtained in the step 203;
step 205, the original signal is represented as the sum of IMF components and residual classifications obtained by decomposition in step 204. The extreme point continuation and the segmentation three-time Hermite interpolation are adopted to improve the empirical mode decomposition, reduce the average times of the empirical mode decomposition set and solve the endpoint distortion phenomenon existing in the mode decomposition.
Further, the white noise signal in the step 201 is not higher than 0.3 times of the root mean square of the original signal.
Further, the parameter selection of optimizing the depth confidence network by adopting the pigeon swarm optimization algorithm is to optimize the positions and speeds of the nodes of the BP neural network and the limited Boltzmann machine coded to pigeon swarm individuals.
Further, the fitness value of the pigeon optimization algorithm is the mean square error of the deep confidence network energy consumption prediction model.
Further, the limited boltzmann machine is provided with a three-layer stack.
The workshop energy consumption management system based on the improved deep confidence network comprises a data acquisition module and an energy consumption prediction module, wherein the data acquisition module is used for acquiring energy consumption data of workshop equipment, forming a data set and sending the data set to the energy consumption prediction module, and the energy consumption prediction module is used for realizing a workshop energy consumption prediction method based on the improved deep confidence network.
Further, the device monitoring system comprises a device monitoring module and a management center, wherein the device monitoring module receives the energy consumption data and the running state information of the workshop device acquired by the data acquisition module, judges whether the device is abnormal according to whether the state range value is exceeded, and sends the information of whether the device monitoring module is abnormal to the management center.
Compared with the prior art, the invention has the advantages that:
1. the method combines the integrated empirical mode decomposition, the sample entropy, the pigeon cluster optimization algorithm and the deep confidence network, establishes the workshop energy consumption prediction model, improves the accuracy of workshop energy consumption prediction, and has higher accuracy of energy consumption prediction compared with the traditional workshop energy consumption prediction method.
2. The method adopts the parameter selection of the pigeon optimization algorithm to the deep confidence network, and forms the weight of each layer of the neural network and the node number of the hidden layer into the position and the speed of the pigeon individual, thereby reducing the calculation amount of algorithm optimization and effectively improving the convergence speed of the algorithm.
3. The workshop energy consumption management system provided by the invention realizes the prediction of the workshop energy consumption, obtains the running state through the equipment monitoring module, and can timely process the equipment energy consumption under the condition of large fluctuation of the equipment energy consumption by comparing the running state with the state range value, thereby quickly reducing the equipment energy consumption and improving the management efficiency of the workshop energy consumption.
Drawings
FIG. 1 is a schematic diagram of a plant energy consumption management system module based on an improved deep belief network.
FIG. 2 is a flow chart of a plant energy consumption prediction method based on an improved deep belief network.
FIG. 3 is a schematic diagram of a process for optimizing deep belief network parameters using a pigeon swarm optimization algorithm.
Detailed Description
The invention is further illustrated, but is not limited, by the following examples.
Referring to fig. 1, the plant energy consumption management system based on the improved deep belief network includes a data acquisition module 100, an energy consumption prediction module 200, a management center 300, and a device monitoring module 400. In this embodiment, the data acquisition module 100 is mainly configured to acquire energy consumption data of workshop equipment in real time, and at the same time, upload the acquired data to the energy consumption prediction module 200 for analysis and processing. The specific operation is as follows: taking an hour as an acquisition period of workshop equipment energy consumption data, acquiring workshop energy consumption data once every 5 minutes, uploading the workshop energy consumption data once every 1 hour, preprocessing the acquired data by the data acquisition module 100, analyzing the energy consumption data by the energy consumption prediction module 200, establishing a graph of change of workshop energy consumption with time, and simultaneously storing the preprocessed workshop energy consumption data into a MySQL database of the energy consumption management system cloud platform 500.
The data acquisition module 100 includes sensors mounted on the plant equipment, specifically for acquiring energy consumption data and operating status information of different types of plant equipment. The sensor and the equipment monitoring module 400 in the energy consumption management system of the invention are connected in a distributed manner through the node of the Internet of things. The energy consumption prediction module 200 adopts the workshop energy consumption prediction method of the improved deep confidence network provided by the invention, and has the main functions of analyzing the workshop energy consumption data, establishing a workshop energy consumption prediction model, predicting the future trend of the workshop energy consumption according to the historical data of the workshop energy consumption, and simultaneously transmitting the prediction result to the management center 300 and the cloud platform 500 for workshop equipment management.
The energy consumption data and the running state information collected by the sensor are also sent to the device monitoring module 400, meanwhile, the device monitoring module 400 performs preprocessing on the state information collected by the current device, and then compares the preprocessed running state information of the device with the running state range of the device stored in the cloud platform 500 to determine whether the current running device is in a safe state, and if the current running device exceeds the state range value, the device is in an abnormal state.
The device monitoring module 400 sends the operation state of the workshop device to the management center 300 of the energy consumption management system, and the management center 300 sends the operation state of the workshop device to workshop staff, so that the workshop staff can process the workshop operation safety risk in time.
The management center 300 functions mainly include three parts: workshop equipment management, workshop personnel management and early warning management. The workshop equipment management is mainly to discover the operation safety risk of the processing workshop equipment in time through the equipment monitoring module 400. The workshop personnel management comprises system personnel management and personnel authority management, wherein the system personnel management is mainly responsible for adding and deleting the information of the energy consumption management system personnel, and meanwhile is responsible for managing the account information of the personnel, and the personnel authority management is responsible for setting different user authorities for system staff. The early warning management is mainly responsible for receiving equipment abnormality information transmitted from equipment state monitoring and equipment energy consumption monitoring in the equipment monitoring module 400, and coping with different fault information of workshop equipment according to equipment state abnormality and equipment energy consumption abnormality processing methods in the energy consumption management system, and transmitting the fault information of the equipment to workshop staff, so that the workshop staff can conveniently overhaul and maintain the equipment, and the running cost of enterprises is reduced.
In addition, it is required to separately explain that different types of equipment sensors are installed on different workshop equipment, the equipment sensors comprise a temperature sensor, a noise sensor, a voltage sensor, a current sensor and the like, and the current equipment state information to be collected comprises information such as the current temperature, noise and vibration of the workshop equipment.
The plant energy consumption prediction method based on the improved deep belief network adopted by the energy consumption prediction module 200 in the plant energy consumption management system specifically includes the following steps, please refer to fig. 2.
Step 1, information transmitted by the data acquisition module 100 is received to acquire energy consumption data of workshop equipment, and a data set is formed.
Step 2, normalizing the data set, wherein the normalization process normalizes the data to [0,1] according to the following formula:
wherein X is max Represents the maximum value in the dataset, X min Representing minimum value in data set, X nor Representing the normalized value.
And carrying out empirical mode decomposition on the normalized data sets and reconstructing based on sample entropy, so as to reduce randomness and fluctuation among different data sets. Empirical mode decomposition cannot completely solve the problem of mode aliasing, and a problem arises in that the number of times of adding white noise and a false component are excessive, resulting in an increase in the calculation amount of the decomposition. The invention adopts extreme point continuation and segmentation three-time Hermite interpolation to improve empirical mode decomposition, and comprises the following specific steps:
step 201, firstly adding a white noise signal which is not higher than 0.3 times of the root mean square of an original signal into a signal to be decomposed;
step 202, generating an original signal x (t) AddingTwo groups of positive and negative white noise signals w with equal absolute values +(t) And w -(t) Then the original signal is decomposed as follows
Wherein d i，j(t) Representing the ith IMF component, p, obtained by the jth decomposition j(t) Representing the residual component obtained by the jth decomposition;representing the ith IMF component resulting from the decomposition, < >>Representing the residual component resulting from the decomposition;
step 203, averaging the two groups of IMF components obtained in step 202 so as to eliminate the residual signal of white noise to the greatest extent;
step 204, for the average value d obtained in step 203 i(t) Performing EMD decomposition again;
wherein, c 1(t) Decomposition of the 1 st IMF component, where q 1(t) Representing the first residual component, c k(t) Represents the k-th IMF component, q, obtained by decomposition k(t) Represents the residue after decomposition, k=1, 2,..m.
Step 205 the original signal is represented as the sum of IMF components and residual classifications obtained by decomposition in step 204,
wherein r is (t) Representing the residual component resulting from the final decomposition.
Since more IMF components are generated by empirical mode decomposition, if the IMF components are directly subjected to energy consumption prediction, the calculation scale and time of the energy consumption prediction model are increased. In order to reduce the calculation amount of an energy consumption prediction model, the invention calculates the Sample Entropy (SE) value of each component on the basis of empirical mode decomposition, and reconstructs a sequence with the SE value close to the Sample Entropy (SE) value into a new sequence, wherein the sequence is formed by sequentially arranging data obtained by substituting calculation formulas of the SE values after parameters, and the parameters are formed by energy consumption data parameters obtained according to actual testing of a factory and set defined parameter dimensions m and threshold r. And establishing a workshop energy consumption prediction model for the new sequence, wherein the method comprises the following specific steps of:
step 206, assuming that the plant energy consumption data is a set of m-dimensional time series x (t), i.e. x m (1)，...，x m (N-m+1), wherein x m (i) = { x (i), x (i+1),...
Step 207, set x m (i) And x m (j) Absolute distance d [ x ] between m (i)，x m (j)]The method comprises the following steps:
step 208, setting a threshold r, and counting d [ x ] for each i m (i)，x m (j)]The number < r, denoted N m(i) . Definition of the definitionAt the same time calculate the mean->
Step 209, m=m+1, repeating steps 207, 208 to obtain
Step 210, the SE of the sequence x (t) is:
dividing the data processed in the step 2 into a training set and a testing set according to the actual factory, wherein the training set is used for training the workshop energy consumption prediction model, and the testing set is used for testing the accuracy of the workshop energy consumption prediction model. The method specifically comprises the steps 3 and 4.
And 3, establishing a pigeon swarm optimization algorithm optimized deep confidence network energy consumption prediction model, wherein the deep confidence network energy consumption prediction model comprises a three-layer limited Boltzmann machine stack and a BP neural network, the three-layer limited Boltzmann machine stack performs layer-by-layer characteristic value self-extraction on an input sequence through unsupervised learning, the extracted characteristic value is used as the BP neural network input into the top layer, the BP neural network trains and fits the characteristic value and output power of data to complete regression prediction, and meanwhile, the BP neural network adjusts all network parameters of the limited Boltzmann machine at the bottom layer. The deep confidence network energy consumption prediction model optimizes parameter selection of the deep confidence network by adopting a pigeon cluster optimization algorithm, and the process comprises the following steps of:
step 301, initializing relevant parameters of a deep confidence network and a pigeon swarm algorithm.
Step 302, encoding weights of the neural network and node numbers of hidden layers of three limited Boltzmann machines to positions x of pigeon flock individuals i And velocity v i The location and velocity of individual pigeon flocks are initialized.
Step 303, selecting a Mean Square Error (MSE) of a deep confidence network energy consumption prediction model as an adaptability value of a pigeon cluster optimization algorithm, calculating the adaptability value, and simultaneously recording an optimal position and a global optimal position of an individual pigeon cluster, wherein the calculation formula of the MSE is as follows:
where Yi is the real data of the data,for prediction data, n is the number of samples.
Step 304, updating the position and speed of the pigeon crowd individuals according to the following formula;
wherein R is a map and compass factor, and is set to be a value between 0 and 1, and x gbest For the global optimal position, t represents the current iteration number, and rand is [0,1]Random numbers in between.
Step 305, if the guidance factor meets the termination condition, then step 306 is executed, otherwise, step 303 is executed again.
Step 306, sorting pigeon flock individuals according to the size of the fitness value, and deleting the individuals with the fitness value in the second half.
Step 307, finding the central position x of the pigeon cluster by center (t)；
In the middle ofRepresenting the central position of the pigeon flock.
Step 308, updating the individual optimal position and the global optimal position.
309, increasing the cooperation capability among different individuals by utilizing the inter-pigeon group communication, and updating the positions of the pigeon group individuals according to the following formula;
in the middle ofRepresenting the new position of the pigeon after learning, +.>Indicating the current position of the pigeon.
Step 310, updating the individual optimal position and the global optimal position.
Step 311, calculating fitness value changes due to location update: Δf=f i (x i (t+1))-f i (x i (T)) if Δf < 0 or exp (- Δf/T) > rand, T is temperature, then the position is updated, otherwise the original position is preserved.
Step 312, judging whether the landmark operator meets the termination condition, if yes, outputting an optimal parameter value of the deep confidence network, otherwise, executing a cooling operation, wherein t=t.decalycals, which is an annealing coefficient, and returning to step 307.
And 4, testing the trained deep belief network energy consumption prediction model by using test data, and predicting workshop energy consumption by using the tested deep belief network energy consumption prediction model.
It should be noted that the specific methods of the above-described embodiments may be stepped to form computer program products that may be stored on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.). In addition, the application can be implemented in hardware, software or a combination of hardware and software, or can be a computer device comprising at least one processor and a memory, and the energy consumption prediction module of the plant energy consumption management system based on the improved deep belief network can be the computer device or the computer program product. When the energy consumption prediction module is a computer device, the memory stores a computer program for implementing the flow steps of the workshop energy consumption prediction method based on the improved deep belief network, and the processor is used for executing the computer program on the memory to form the workshop energy consumption prediction method based on the improved deep belief network in the embodiment.
Claims (5)
1. The workshop energy consumption prediction method based on the improved deep belief network is characterized by comprising the following steps of:
step 1, collecting energy consumption data of workshop equipment to form a data set;
step 2, decomposing the data of the data sets by adopting an empirical mode and reconstructing based on sample entropy, so as to reduce randomness and fluctuation among different data sets;
step 3, establishing a deep confidence network energy consumption prediction model optimized by a pigeon cluster optimization algorithm, optimizing parameter selection of the deep confidence network by the deep confidence network energy consumption prediction model by adopting the pigeon cluster optimization algorithm, and training the deep confidence network energy consumption prediction model by the data set processed in the step 2;
step 4, predicting workshop energy consumption by the trained deep confidence network energy consumption prediction model;
wherein, in the step 2, the empirical mode decomposition of the data set includes the following steps:
step 201, adding a white noise signal to an original signal to be decomposed;
step 202, adding two groups of positive and negative white noise signals w with equal absolute values to the original signal +(t) And w -(t) Then decomposing according to the following formula
Wherein d i,j(t) Representing the ith IMF component, p, obtained by the jth decomposition j(t) Representing the residual component obtained by the jth decomposition;representing the ith IMF component resulting from the decomposition, < >>Representing the residual component resulting from the decomposition;
step 203, calculating the average value of the two groups of IMF components obtained in the step 202;
step 204, performing EMD decomposition on the average value obtained in the step 203;
step 205, representing the original signal as the sum of IMF components and residual classifications obtained by decomposition in step 204;
the deep confidence network energy consumption prediction model comprises a three-layer limited Boltzmann machine stack and a BP neural network, wherein the three-layer limited Boltzmann machine stack performs layer-by-layer characteristic value self-extraction on an input sequence through unsupervised learning, the extracted characteristic value is used as the BP neural network input into a top layer, the BP neural network trains and fits the characteristic value and output power of data to complete regression prediction, and meanwhile, the BP neural network adjusts all network parameters of a limited Boltzmann machine at the bottom layer; the parameter selection of optimizing the deep confidence network by adopting the pigeon optimization algorithm is to encode the weight of the BP neural network and the node number of the limited Boltzmann machine to the position and the speed of pigeon individuals for optimization, and the adaptability value of the pigeon optimization algorithm is the mean square error of the energy consumption prediction model of the deep confidence network.
2. The improved deep belief network based plant energy consumption prediction method of claim 1, wherein the dataset is normalized prior to employing empirical mode decomposition and sample entropy based reconstruction, the normalization normalizing data to [0,1] according to the following formula:
wherein X is max Represents the maximum value in the dataset, X min Representing minimum value in data set, X nor Representing the normalized value.
3. The improved deep belief network based plant energy consumption prediction method according to claim 1, wherein the white noise signal is not higher than 0.3 times the root mean square of the original signal in the step 201.
4. A plant energy consumption management system based on an improved deep belief network, comprising a data acquisition module and an energy consumption prediction module, wherein the data acquisition module is used for acquiring energy consumption data of plant equipment, forming a data set and sending the data set to the energy consumption prediction module, and the energy consumption prediction module is used for realizing the plant energy consumption prediction method based on the improved deep belief network according to any one of claims 1 to 3.
5. The plant energy consumption management system based on the improved deep belief network according to claim 4, comprising a plant monitoring module and a management center, wherein the plant monitoring module receives the energy consumption data and the operation state information of the plant equipment collected by the data collection module and judges whether the equipment is abnormal according to whether the state range value is exceeded, and the plant monitoring module sends the information of whether the equipment is abnormal to the management center.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310001073.1A CN116070669B (en) | 2023-01-03 | 2023-01-03 | Workshop energy consumption prediction method and management system based on improved deep belief network |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202310001073.1A CN116070669B (en) | 2023-01-03 | 2023-01-03 | Workshop energy consumption prediction method and management system based on improved deep belief network |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN116070669A CN116070669A (en) | 2023-05-05 |
| CN116070669B true CN116070669B (en) | 2024-01-26 |
Family
ID=86178059
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202310001073.1A Active CN116070669B (en) | 2023-01-03 | 2023-01-03 | Workshop energy consumption prediction method and management system based on improved deep belief network |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN116070669B (en) |
Families Citing this family (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116647819B (en) * | 2023-07-27 | 2023-11-07 | 深圳市中科智联有限公司 | Instrument energy consumption monitoring method and system based on sensor network |
Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109391515A (en) * | 2018-11-07 | 2019-02-26 | 武汉烽火技术服务有限公司 | Network failure prediction technique and system based on dove group's algorithm optimization support vector machines |
| CN112364291A (en) * | 2020-11-17 | 2021-02-12 | 哈工大机器人（合肥）国际创新研究院 | Pre-filtering extreme point optimization set empirical mode decomposition method and device |
| CN114997044A (en) * | 2022-05-17 | 2022-09-02 | 电子科技大学 | Energy consumption optimization method and data acquisition system for central air conditioner |
| CN115169218A (en) * | 2022-05-30 | 2022-10-11 | 上海大学 | Gear vibration noise estimation method based on deep belief network |
| CN115456135A (en) * | 2022-08-29 | 2022-12-09 | 中铁十二局集团有限公司 | Track bed deviation prediction method based on improved GWO-DBN rail inspection vehicle detection data |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN107220734B (en) * | 2017-06-26 | 2020-05-12 | 江南大学 | Numerical control lathe turning process energy consumption prediction system based on decision tree |
-
2023
- 2023-01-03 CN CN202310001073.1A patent/CN116070669B/en active Active
Patent Citations (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN109391515A (en) * | 2018-11-07 | 2019-02-26 | 武汉烽火技术服务有限公司 | Network failure prediction technique and system based on dove group's algorithm optimization support vector machines |
| CN112364291A (en) * | 2020-11-17 | 2021-02-12 | 哈工大机器人（合肥）国际创新研究院 | Pre-filtering extreme point optimization set empirical mode decomposition method and device |
| CN114997044A (en) * | 2022-05-17 | 2022-09-02 | 电子科技大学 | Energy consumption optimization method and data acquisition system for central air conditioner |
| CN115169218A (en) * | 2022-05-30 | 2022-10-11 | 上海大学 | Gear vibration noise estimation method based on deep belief network |
| CN115456135A (en) * | 2022-08-29 | 2022-12-09 | 中铁十二局集团有限公司 | Track bed deviation prediction method based on improved GWO-DBN rail inspection vehicle detection data |
Non-Patent Citations (3)
| Title |
|---|
| Fine-tuned artificial intelligence model using pigeon optimizer for prediction of residual stresses during turning of Inconel 718;Ammar H. Elsheikh 等;《Journal of Materials Research and Technology》;第15卷;3622-3634 * |
| 基于鸽群优化算法的支持向量机在电力需求总量预测中的应用;田书欣等;《电力自动化设备》(第5期);173-181 * |
| 改进鲸鱼算法构建混合模型的建筑能耗预测;王茹;宋爽;贺佳;;计算机测量与控制(02);197-201+205 * |
Also Published As
| Publication number | Publication date |
|---|---|
| CN116070669A (en) | 2023-05-05 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN105117602B (en) | A kind of metering device running status method for early warning | |
| CN106529719B (en) | Wind power prediction method based on particle swarm optimization algorithm wind speed fusion | |
| CN112488395A (en) | Power distribution network line loss prediction method and system | |
| CN111275288A (en) | XGboost-based multi-dimensional data anomaly detection method and device | |
| CN110222991B (en) | Metering device fault diagnosis method based on RF-GBDT | |
| CN110717610B (en) | Wind power prediction method based on data mining | |
| CN112818604A (en) | Wind turbine generator risk degree assessment method based on wind power prediction | |
| CN116070669B (en) | Workshop energy consumption prediction method and management system based on improved deep belief network | |
| CN116150897A (en) | Machine tool spindle performance evaluation method and system based on digital twin | |
| CN111461564A (en) | Wind turbine generator power characteristic evaluation method based on cloud model and optimal combined weighting | |
| CN109299208B (en) | Intelligent visual risk assessment method for transmission tower in typhoon disaster | |
| CN116380445B (en) | Equipment state diagnosis method and related device based on vibration waveform | |
| CN112884243A (en) | Air quality analysis and prediction method based on deep learning and Bayesian model | |
| CN115034485A (en) | Wind power interval prediction method and device based on data space | |
| CN112746934A (en) | Method for diagnosing fan fault through self-association neural network | |
| CN114322199A (en) | Ventilation system autonomous optimization operation regulation and control platform and method based on digital twins | |
| CN115438726A (en) | Device life and fault type prediction method and system based on digital twin technology | |
| CN112784920A (en) | Cloud-side-end-coordinated dual-anti-domain self-adaptive fault diagnosis method for rotating part | |
| CN116029435A (en) | Environmental comfort early warning system is bred to live pig facility | |
| CN112819238A (en) | Short-term wind power prediction method based on chaotic chicken flock optimization algorithm | |
| CN115271225A (en) | Wind power-wind power modeling method based on wavelet denoising and neural network | |
| CN113689072B (en) | Marine wind turbine running state evaluation method based on Markov chain | |
| CN112149870B (en) | Pig house ammonia concentration combination prediction method based on ISODATA clustering and Elman neural network | |
| CN116797204A (en) | Primary air quantity fault early warning method for coal mill based on wavelet decomposition and reconstruction and TCN-GRU-Self-Attention | |
| CN116776209A (en) | Method, system, equipment and medium for identifying operation state of gateway metering device |
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 |