CN110232444B - Geological monitoring BP neural network optimization method, device, equipment and storage medium - Google Patents
Geological monitoring BP neural network optimization method, device, equipment and storage medium Download PDFInfo
- Publication number
- CN110232444B CN110232444B CN201910525396.4A CN201910525396A CN110232444B CN 110232444 B CN110232444 B CN 110232444B CN 201910525396 A CN201910525396 A CN 201910525396A CN 110232444 B CN110232444 B CN 110232444B
- Authority
- CN
- China
- Prior art keywords
- neural network
- optimized
- network structure
- algorithm
- initial
- 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
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N33/00—Investigating or analysing materials by specific methods not covered by groups G01N1/00 - G01N31/00
- G01N33/24—Earth materials
-
- 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
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/12—Computing arrangements based on biological models using genetic models
- G06N3/126—Evolutionary algorithms, e.g. genetic algorithms or genetic programming
-
- 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"
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- Life Sciences & Earth Sciences (AREA)
- Health & Medical Sciences (AREA)
- Theoretical Computer Science (AREA)
- Biophysics (AREA)
- General Physics & Mathematics (AREA)
- Business, Economics & Management (AREA)
- General Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Strategic Management (AREA)
- Chemical & Material Sciences (AREA)
- Computational Linguistics (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Evolutionary Biology (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Bioinformatics & Computational Biology (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Human Resources & Organizations (AREA)
- Economics (AREA)
- General Life Sciences & Earth Sciences (AREA)
- Marketing (AREA)
- Game Theory and Decision Science (AREA)
- Remote Sensing (AREA)
- Geology (AREA)
- Operations Research (AREA)
- Quality & Reliability (AREA)
- Tourism & Hospitality (AREA)
- General Business, Economics & Management (AREA)
- Environmental & Geological Engineering (AREA)
- Physiology (AREA)
- Development Economics (AREA)
- Entrepreneurship & Innovation (AREA)
- Genetics & Genomics (AREA)
Abstract
The invention relates to the technical field of artificial intelligence, and discloses a geological monitoring BP neural network optimization method, device, equipment and storage medium. When the geological monitoring BP neural network to be optimized is optimized, local and global search is realized by combining a hill climbing algorithm, a genetic algorithm and a simulated annealing algorithm with the BP neural network, so that compared with the BP neural network which is optimized by only utilizing the genetic algorithm, the optimized BP neural network can effectively improve global convergence and avoid trapping in a local extremum.
Description
Technical Field
The invention relates to the technical field of artificial intelligence, in particular to a geological monitoring BP neural network optimization method, device, equipment and storage medium.
Background
The genetic algorithm is a highly parallel, random and self-adaptive search algorithm evolved by using an evolution mechanism of survival, excellence and decline of a suitable person in the natural world as a reference, and is widely applied to the fields of machine learning, artificial intelligence and the like at present.
A Back Propagation Neural Network (BP Neural Network) is an intelligent algorithm provided by simulating the Neural system structure of human brain, and is trained according to an error Back Propagation algorithm, input information is calculated layer by layer in the forward Propagation process and transmitted to an output layer to obtain actual output, if the error between the actual output and expected output fails to meet requirements, the actual output is transmitted to a Back Propagation process, weight and threshold of the Network are modified layer by layer through the error and transmitted to an input layer, and optimal weight and threshold are obtained through repeated training of the forward and Back Propagation processes so as to achieve the prediction target of the Network.
At present, in order to overcome the defects that the BP neural network has a low convergence rate and is prone to trapping local minimum points, a genetic algorithm is usually combined with the BP neural network to solve the technical problems. However, the genetic algorithm still has the problems that the initial population is distributed in a local area, the search range is limited, the local convergence is poor, the diversity is reduced in the later iteration stage, and the like, so the existing method for optimizing the BP neural network by simply utilizing the genetic algorithm is not ideal.
The above is only for the purpose of assisting understanding of the technical solution of the present invention, and does not represent an admission that the above is the prior art.
Disclosure of Invention
The invention mainly aims to provide a geological monitoring BP neural network optimization method, a geological monitoring BP neural network optimization device, geological monitoring BP neural network optimization equipment and a storage medium, and aims to solve the technical problems.
In order to achieve the purpose, the invention provides an optimization method of a geological monitoring BP neural network, which comprises the following steps:
determining an initial network structure of a geological monitoring BP neural network to be optimized;
global optimization is carried out on the initial weight and the initial threshold of the initial network structure by adopting a hill climbing algorithm, a genetic algorithm and a simulated annealing algorithm to obtain an optimized weight and an optimized threshold;
acquiring training data, wherein the training data comprises longitude information, latitude information, elevation information and affiliated functional area information of the land with known monitoring results;
performing iterative training on the initial network structure according to the longitude information, the latitude information, the altitude information and the functional area information, and modifying the optimal weight and the optimal threshold in the training process until the iteration times reach a preset maximum iteration time or an iteration stopping condition is met;
and taking the network structure when the training is stopped as the optimized network structure of the geological monitoring BP neural network to be optimized.
Preferably, the step of determining an initial network structure of the geological monitoring BP neural network to be optimized includes:
determining the neuron number of an input layer and the neuron number of an output layer of the geological monitoring BP neural network to be optimized;
determining the value range of the neuron number of the hidden layer of the geological monitoring BP neural network to be optimized according to the neuron number of the input layer and the neuron number of the output layer;
acquiring the minimum mean square error of each value in the value range, and taking the neuron number corresponding to the minimum mean square error as the neuron number of the hidden layer;
and determining the initial network structure according to the neuron number of the hidden layer.
Preferably, the step of performing global optimization on the initial weight and the initial threshold of the initial network structure by using a hill-climbing algorithm, a genetic algorithm and a simulated annealing algorithm to obtain an optimized weight and an optimized threshold includes:
performing real number coding on each neuron in the initial network structure by adopting the genetic algorithm to obtain a population data set corresponding to the initial network structure;
performing population division on the population data set by adopting the hill climbing algorithm to obtain a large population and a small population;
local optimization is carried out on the small population by adopting the hill climbing algorithm and the genetic algorithm, and the found local optimal weight and the local optimal threshold are assigned to the large population;
and performing global optimization on the large population by adopting the genetic algorithm and the simulated annealing algorithm, and taking the searched global optimal weight and global optimal threshold as the optimal weight and optimal threshold of the initial network structure.
Preferably, the step of locally optimizing the small population by using the hill-climbing algorithm and the genetic algorithm comprises:
setting a first predicted value and a first expected value for the small population;
calculating a first fitness value of each neuron in the small population according to the first predicted value, the first expected value and a first fitness function;
repeatedly performing selection, crossing and mutation genetic operations on the first fitness value of each neuron in the small population by adopting the genetic algorithm;
and local optimization is carried out on the small population after each genetic operation by adopting the hill climbing algorithm until the local optimal weight and the local optimal threshold are obtained.
Preferably, the step of global optimization of the large population by using the genetic algorithm and the simulated annealing algorithm comprises:
setting a second predicted value and a second expected value for the large population;
calculating a second fitness value of each neuron in the large population according to the second predicted value, the second expected value and a second fitness function;
repeatedly performing selection, crossing and mutation genetic operations on the second fitness value of each neuron in the large population by adopting the genetic algorithm;
and carrying out global optimization on the large population after each genetic operation by adopting a simulated annealing algorithm until the global optimal weight and the global optimal threshold are obtained.
Preferably, the step of acquiring training data includes:
and acquiring sample data, and performing maximum and minimum normalization processing on the sample data to obtain the training data.
Preferably, the step of iteratively training the initial network structure according to the longitude information, the latitude information, the altitude information, and the functional area information includes:
and inputting the longitude information, the latitude information, the altitude information and the functional area information as input parameters into the initial network structure, and repeatedly carrying out forward propagation and backward propagation training operations on the initial network structure.
In addition, in order to achieve the above object, the present invention further provides an optimization apparatus for geological monitoring BP neural network, the apparatus comprising:
the system comprises a first determination module, a second determination module and a third determination module, wherein the first determination module is used for determining an initial network structure of a geological monitoring BP neural network to be optimized;
the optimization module is used for carrying out global evolution optimization on the initial weight and the initial threshold of the initial network structure by adopting a hill-climbing algorithm, a genetic algorithm and a simulated annealing algorithm to obtain an optimized weight and an optimized threshold;
the acquisition module is used for acquiring training data, wherein the training data comprises longitude information, latitude information and altitude information of the land with known monitoring results and the information of the affiliated functional area;
the training module is used for carrying out iterative training on the initial network structure according to the longitude information, the latitude information, the altitude information and the functional area information, and modifying the optimal weight and the optimal threshold in the training process until the iteration times reach a preset maximum iteration time or the iteration stopping condition is met;
and the second determining module is used for taking the network structure when the training is stopped as the optimized network structure of the geological monitoring BP neural network to be optimized.
In addition, in order to achieve the above object, the present invention further provides an optimization apparatus for geological monitoring BP neural network, including: the optimization program of the geological monitoring BP neural network is configured to realize the steps of the optimization method of the geological monitoring BP neural network.
In addition, to achieve the above object, the present invention further provides a computer readable storage medium, which stores thereon an optimization program of the geological monitoring BP neural network, wherein the optimization program of the geological monitoring BP neural network, when executed by a processor, implements the steps of the optimization method of the geological monitoring BP neural network as described above.
According to the optimization scheme of the geological monitoring BP neural network, when the geological monitoring BP neural network to be optimized is optimized, local and global search is realized by combining a hill climbing algorithm, a genetic algorithm and a simulated annealing algorithm with the BP neural network, so that compared with the BP neural network which is optimized by only utilizing the genetic algorithm, the optimized BP neural network can effectively improve global convergence and avoid falling into local extremum.
Drawings
FIG. 1 is a schematic structural diagram of an optimization device of a geological monitoring BP neural network of a hardware operating environment according to an embodiment of the invention;
FIG. 2 is a schematic flow chart of a first embodiment of the optimization method of the geological monitoring BP neural network according to the invention;
FIG. 3 is a schematic flow chart of a second embodiment of the optimization method of the geological monitoring BP neural network according to the invention;
fig. 4 is a block diagram of the first embodiment of the optimization apparatus for geological monitoring BP neural network according to the present invention.
The implementation, functional features and advantages of the objects of the present invention will be further explained with reference to the accompanying drawings.
Detailed Description
It should be understood that the specific embodiments described herein are merely illustrative of the invention and do not limit the invention.
Referring to fig. 1, fig. 1 is a schematic structural diagram of an optimization device of a geological monitoring BP neural network in a hardware operating environment according to an embodiment of the present invention.
As shown in fig. 1, the optimization device of the geological monitoring BP neural network may include: a processor 1001, such as a Central Processing Unit (CPU), a communication bus 1002, a user interface 1003, a network interface 1004, and a memory 1005. Wherein a communication bus 1002 is used to enable connective communication between these components. The user interface 1003 may include a Display screen (Display), an input unit such as a Keyboard (Keyboard), and the optional user interface 1003 may also include a standard wired interface, a wireless interface. The network interface 1004 may optionally include a standard wired interface, a WIreless interface (e.g., a WIreless-FIdelity (WI-FI) interface). The Memory 1005 may be a Random Access Memory (RAM) Memory, or may be a Non-Volatile Memory (NVM), such as a disk Memory. The memory 1005 may alternatively be a storage device separate from the processor 1001.
Those skilled in the art will appreciate that the architecture shown in FIG. 1 does not constitute a limitation of the optimization apparatus for geological monitoring of BP neural networks, and may include more or fewer components than shown, or some components in combination, or a different arrangement of components.
As shown in fig. 1, the memory 1005, which is a storage medium, may include therein an operating system, a network communication module, a user interface module, and an optimization program of the geological monitoring BP neural network.
In the optimization device of the geological monitoring BP neural network shown in FIG. 1, the network interface 1004 is mainly used for data communication with a network server; the user interface 1003 is mainly used for data interaction with a user; the processor 1001 and the memory 1005 in the optimization device of the geological monitoring BP neural network may be arranged in the optimization device of the geological monitoring BP neural network, and the optimization device of the geological monitoring BP neural network calls the optimization program of the geological monitoring BP neural network stored in the memory 1005 through the processor 1001 and executes the optimization method of the geological monitoring BP neural network provided by the embodiment of the present invention.
The embodiment of the invention provides an optimization method of a geological monitoring BP neural network, and referring to FIG. 2, FIG. 2 is a flow diagram of a first embodiment of the optimization method of the geological monitoring BP neural network.
In this embodiment, the optimization method of the geological monitoring BP neural network includes the following steps:
and S10, determining an initial network structure of the geological monitoring BP neural network to be optimized.
It should be understood that the so-called BP Neural Network, known as Back Propagation Neural Network, is a multi-layer feedforward Neural Network trained according to an error Back Propagation algorithm.
Correspondingly, the geological monitoring BP neural network in this embodiment is a BP neural network for monitoring geological information of soil to be monitored.
Furthermore, it should be understood that since the network structure of the BP neural network generally includes three major layers, respectively: an input layer, a hidden layer (also called intermediate layer) and an output layer.
The number of layers of the hidden layer and the number of neurons in each hidden layer may be determined according to the data size to be analyzed and processed by the BP neural network, and will not be described herein again.
Accordingly, the initial network structure determined in this embodiment may also be roughly divided into an input layer, an implicit layer, and an output layer.
Further, in order to ensure that the geological information of the soil to be monitored can be better predicted by the geological monitoring BP neural network obtained through subsequent optimization, when the initial network structure is determined, parameters such as an activation function, a learning rate, a training function, a performance function and a motion term coefficient corresponding to the initial network structure can be determined.
Specifically, in practical application, the training function is determined according to a training method selected in the subsequent iterative training of the initial network structure.
For example, when a momentum gradient descent method is used as a training method for the initial network structure, the training function is a momentum gradient descent (slingdm) function.
In addition, in a specific application, the performance function may specifically adopt a Mean Square Error (MES) function.
The parameters such as the activation function, the learning rate, the motion coefficient and the like can be selected and determined according to actual conditions.
It should be understood that the above is only an example, and the technical solution of the present invention is not limited in any way, and in a specific application, a person skilled in the art may set the technical solution as needed, and the present invention is not limited thereto.
In addition, it should be noted that the execution main body of the embodiment may be any terminal device with a processing function, such as a computer, a tablet computer, a mobile phone, and the like, which are not listed here any more, and no limitation is made to this.
And S20, carrying out global optimization on the initial weight and the initial threshold of the initial network structure by adopting a hill-climbing algorithm, a genetic algorithm and a simulated annealing algorithm to obtain an optimized weight and an optimized threshold.
Specifically, in the embodiment, a hill climbing algorithm and a genetic algorithm are combined, local optimization is performed on an initial weight and an initial threshold of the initial network structure, then the genetic algorithm and a simulated annealing algorithm are combined, global optimization is performed on the initial network structure by using a local optimal weight and a local optimal threshold obtained through local optimization, and then an optimized weight and an optimized threshold corresponding to the initial network structure are obtained.
In order to facilitate understanding of the above operations for obtaining the optimized weight and the optimized threshold, the following specific description is made:
(1) And carrying out real number coding on each neuron in the initial network structure by adopting the genetic algorithm to obtain a population data set corresponding to the initial network structure.
Specifically, in genetic algorithms, coding is the key to translating each potential solution to the problem into an individual (chromosome). The whole is converted into individuals, so that the subsequent division of the populations with different sizes is facilitated, and the accuracy of the final optimization result is improved.
Therefore, in this embodiment, before the optimization operation is performed on the initial network structure, a genetic algorithm is used to perform real number encoding on each neuron in the initial network structure, so as to obtain a population data set corresponding to the initial network structure, so that the subsequent operation can be performed conveniently.
In addition, in practical applications, a person skilled in the art may select a binary code, an integer/letter permutation code, or a general data structure code to encode each neuron in the initial network structure as needed, which is not limited herein.
(2) And performing population division on the population data set by adopting the hill climbing algorithm to obtain a large population and a small population.
It should be understood that the hill climbing algorithm is a preferred method, and mainly utilizes feedback information to generate a solution strategy. In order to ensure the accuracy of an optimization result in subsequent operation, the obtained population data sets need to be randomly divided to obtain a large population; and then selecting other individuals in a preset range around each individual in the large population, and combining to obtain a small population.
That is, the individuals in the large population obtained by the first division are relatively dispersed, and thus the gap is large. And the individuals in the small population obtained by dividing the large population are relatively concentrated, so that the gaps are small.
(3) And local optimization is carried out on the small population by adopting the hill-climbing algorithm and the genetic algorithm, and the found local optimal weight and local optimal threshold are assigned to the large population.
Regarding the operation of local optimization of the small population by using the hill-climbing algorithm and the genetic algorithm, the specific implementation flow is roughly as follows:
first, a first predicted value and a first expected value are set for the small population.
Specifically, in the practical application of the first prediction rule and the first expected value in this embodiment, a person skilled in the art may set the analysis structure and the expected structure to be achieved according to the related data of the soil to be monitored, which needs to be analyzed subsequently by the geological monitoring BP neural network to be optimized, and according to the estimated analysis structure of the experimental data, which is not limited herein.
Then, a first fitness value of each neuron in the small population is calculated according to the first predicted value, the first expected value and a first fitness function.
Specifically, when the first fitness value is calculated, a first fitness function is specifically adopted to calculate an absolute value of an error between the first predicted value and the first expected value, and then the reciprocal of the obtained absolute value of the error is used as the first fitness value.
And then, repeatedly carrying out selection, crossing and mutation genetic operations on the first fitness value of each neuron in the small population by adopting the genetic algorithm.
Specifically, when the genetic operation of selecting, crossing and mutating the first fitness value of each neuron in the small population is repeatedly performed, the genetic operation of selecting the first fitness value of each neuron by using a roulette selection method, the genetic operation of crossing the first fitness value of each neuron by using a real number crossing method, and the genetic operation of mutating the first fitness value of each neuron by using a basic bit mutation algorithm may be specifically performed.
In addition, in order to make the iteration process of the above genetic operation relatively reasonable, a maximum number of iterations or a condition for stopping the iteration may be set in advance.
Accordingly, when the repeated selection, crossing and mutation genetic operation reaches the maximum iteration number or meets the condition of stopping iteration, the genetic operation is stopped.
And finally, performing local optimization on the small population after each genetic operation by adopting the hill climbing algorithm until the local optimal weight and the local optimal threshold are obtained.
Specifically, when local optimization is performed, the weight and the threshold corresponding to each neuron in the mini-population are traversed, and then any two traversed weights and any two traversed thresholds are compared until the weights and the thresholds corresponding to all neurons in the mini-population are completely compared, so that the local optimal weight and the local optimal threshold can be obtained.
It should be understood that the above is only an example, and the technical solution of the present invention is not limited in any way, and in a specific application, a person skilled in the art may set the technical solution as needed, and the present invention is not limited thereto.
(4) And performing global optimization on the large population by adopting the genetic algorithm and the simulated annealing algorithm, and taking the searched global optimal weight and global optimal threshold as the optimal weight and optimal threshold of the initial network structure.
Regarding the operation of global optimization on the large population by using the genetic algorithm and the simulated annealing algorithm, the specific implementation flow is roughly as follows:
first, a second predicted value and a second expected value are set for the large population.
Specifically, in the practical application, the second prediction rule and the second expected value in this embodiment may be set by a person skilled in the art according to the related data of the soil to be monitored, which needs to be analyzed subsequently by the geological monitoring BP neural network to be optimized, and according to the analysis structure estimated by the experimental data and the expected structure to be achieved, which is not limited herein.
Then, a second fitness value of each neuron in the large population is calculated according to the second predicted value, the second expected value and a second fitness function.
Specifically, the manner of determining the second fitness value is substantially the same as the manner of determining the first fitness value, and is not repeated here.
In addition, in practical applications, the first fitness function and the second fitness function may be the same fitness function or different fitness functions, and are not limited herein.
And then, repeatedly carrying out the genetic operations of selecting, crossing and mutating on the second fitness value of each neuron in the large population by adopting the genetic algorithm.
Specifically, the repeated selection, crossing and mutation of the second fitness value of each neuron in the large population may be substantially the same as the repeated selection, crossing and mutation of the first fitness value of each neuron in the small population, and will not be described herein again.
In addition, in practical applications, genetic operations such as selection, crossover, mutation, etc. for small and large populations may be performed by the same method or different methods, and are not limited herein.
And finally, carrying out global optimization on the large population after each genetic operation by adopting a simulated annealing algorithm until the global optimal weight and the global optimal threshold are obtained.
Specifically, when global optimization is performed, the weight and the threshold corresponding to each neuron in the large population are traversed, and then any two traversed weights and any two traversed thresholds are compared, until the weights and the thresholds corresponding to all neurons in the large population are compared, the global optimal weight and the global optimal threshold can be obtained.
In addition, it is worth mentioning that, in this embodiment, the simulated annealing algorithm is specifically a Metropolis algorithm (a specific markov chain monte carlo), which is mainly used for comparing any two traversed weights and any two thresholds so as to select a global optimal weight and the global optimal threshold.
Further, in the optimization using the Metropolis algorithm, the initial temperature may be set to 100 ℃.
Further, the temperature reduction parameter and the simulated annealing times required in the optimization process can be specifically determined according to the data situation of the relevant test in practice, and are not limited herein.
It should be understood that the above is only an example, and the technical solution of the present invention is not limited in any way, and in a specific application, a person skilled in the art may set the technical solution as needed, and the present invention is not limited thereto.
Through the above description, it is not difficult to find that, because the gaps of the small population are small, the hill climbing algorithm and the genetic algorithm are firstly adopted to carry out local optimization on the small population, the found optimal value is assigned to the large population, then the genetic algorithm and the simulated annealing algorithm are adopted to carry out global optimization on the large population with larger gaps, and the found global optimal weight and the found global optimal threshold are used as the optimal weight and the optimal threshold of the initial network structure, so that the global convergence of the optimized network structure obtained by subsequent training is greatly improved, and the situation that the optimized network structure falls into a local extreme value can be effectively avoided.
And step S30, acquiring training data.
Specifically, the geological monitoring BP neural network is mainly used for monitoring geological information of soil to be monitored, so that the training data needs to include longitude information, latitude information, elevation information and belonging functional area information of the land with known monitoring results.
In addition, in practical applications, the functional area information is specifically divided by those skilled in the art according to the surrounding environment of the soil to be monitored, for example, the functional area information may be divided into a residential area, a commercial area, an agricultural area, an animal husbandry area, and the like, which are not listed here, and are not limited thereto.
In addition, in practical application, the training data may be obtained from sample data of different soils acquired from each big data platform through data crawling software such as a web crawler.
Specifically, after sample data is acquired from a large data platform by using network crawling, the training data is substantially acquired by performing maximum and minimum normalization processing on the sample data.
It should be understood that, in practical applications, not only the accuracy of the optimized geological monitoring BP neural network prediction data is ensured, but also the training process is simplified as much as possible, so as to reduce the occupation of processor resources of the terminal device and increase the calculation speed. Therefore, on the premise of ensuring the accuracy of the training result, the embodiment obtains the training data by carrying out normalization processing on the sample data, thereby greatly simplifying the training process, effectively reducing the occupation of the processor resources of the computer equipment and accelerating the calculation speed.
And S40, performing iterative training on the initial network structure according to the longitude information, the latitude information, the altitude information and the functional area information, and modifying the optimal weight and the optimal threshold in the training process until the iteration times reach a preset maximum iteration time or the iteration stopping condition is met.
Specifically, the iterative training performed on the initial network structure in this embodiment specifically includes: and inputting the longitude information, the latitude information, the altitude information and the functional area information as input parameters into the initial network structure, and repeatedly carrying out forward propagation and backward propagation training operations on the initial network structure.
For ease of understanding, the following is a detailed description of the process of one training operation:
firstly, distributing class labels for training data of known monitoring results;
then, forward training (in a forward propagation manner, namely, input from an input layer, passing through a hidden layer, and finally output from an output layer) the initial network structure by using training data carrying the class labels, namely, the longitude information, the latitude information, the altitude information, and the functional area information, to obtain a training result;
then, comparing the training result with a known monitoring result corresponding to the training data carrying the class label to determine a training error;
and finally, reversely training (in a reverse propagation mode, namely, inputting from an output layer, passing through a hidden layer, and finally outputting from an input layer) the initial network structure by using the training data carrying the class labels and the training errors, and modifying the optimal weight and the optimal threshold according to the training errors in the process of reversely training so as to perfect a training result.
It should be understood that the above is only a specific training mode, and the technical solution of the present invention is not limited in any way, and in a specific application, a person skilled in the art may set the training mode as needed, and the present invention is not limited thereto.
And S50, taking the network structure when the training is stopped as an optimized network node of the geological monitoring BP neural network to be optimized.
Further, in practical application, before stopping performing the iterative training and considering the current network structure as the optimized network structure of the to-be-optimized geological monitoring BP neural network, the current network structure may be tested, and then it is determined whether the current network structure can be regarded as the optimized network structure according to a test result.
Regarding the test procedure, the following can be approximated:
firstly, acquiring initial test data;
then, preprocessing the initial test data (the initial test data needs to be consistent with the preprocessing performed by the sample data so as to ensure the validity of the test result), and obtaining target test data;
then, marking the target test data (specifically, marking the test data of the BP neural network model used for inputting the network structure and the test result corresponding to the test data in the marking process);
secondly, inputting the marked target test data serving as input data into a BP neural network model of the current network structure, and acquiring a test result output by the BP neural network model;
and finally, matching the output test result with the marked test result corresponding to the target test data, if the output test result is matched with the marked test result, considering the current network structure as an optimized network result, and otherwise, continuously executing the iterative training.
It should be understood that the above is only an example, and the technical solution of the present invention is not limited in any way, and in a specific application, a person skilled in the art may set the technical solution as needed, and the present invention is not limited thereto.
It is not difficult to find out through the above description that the optimization method of the geological monitoring BP neural network provided in this embodiment implements local and global search by combining the hill-climbing algorithm, the genetic algorithm, and the simulated annealing algorithm with the BP neural network when the geological monitoring BP neural network to be optimized is optimized, so that compared with the BP neural network optimized by simply using the genetic algorithm, the optimized BP neural network can effectively improve global convergence and avoid falling into a local extremum.
Referring to fig. 3, fig. 3 is a schematic flow chart of a second embodiment of the optimization method for geological monitoring BP neural network according to the present invention.
It should be noted that, in practical application, the prediction accuracy of the geological information of the soil to be monitored by the geological monitoring BP neural network mainly depends on the number of neurons in the hidden layer of the geological monitoring BP neural network, so that the determined initial network structure can be fitted to the reality as much as possible, based on the first embodiment, the optimization method of the geological monitoring BP neural network of this embodiment provides an implementation manner for determining the initial network structure of the geological monitoring BP neural network to be optimized, and details see four substeps included in step S10 in fig. 3:
and a substep S101 of determining the neuron number of the input layer and the neuron number of the output layer of the geological monitoring BP neural network to be optimized.
Specifically, in practical application, the number of neurons in the input layer and the number of neurons in the output layer of the geological monitoring BP neural network to be optimized are determined according to the size of data volume needing to be processed in the later period of the geological monitoring BP neural network to be optimized.
And a substep S102, determining the value range of the neuron number of the hidden layer of the geological monitoring BP neural network to be optimized according to the neuron number of the input layer and the neuron number of the output layer.
And a substep S103, obtaining the minimum mean square error of each value in the value range, and taking the neuron number corresponding to the minimum mean square error as the neuron number of the hidden layer.
It should be understood that in the BP neural network, the selection of the neuron number of the hidden layer is very important, and not only has great influence on the performance of the established BP neural network model, but also is a direct reason for the occurrence of "overfitting" during training.
Therefore, in order to determine the number of neurons in the hidden layer that is actually required to fit, the present embodiment enables the determined number of neurons in the hidden layer to fit the actual measurement requirement as much as possible through the operations given in the substep S102 and the substep S103, so that the geological information of the soil to be monitored can be reasonably predicted by the finally optimized geological monitoring BP neural network, and the accuracy of the prediction result is greatly ensured.
In addition, in practical applications, in order to determine the number of neurons in the hidden layer that fit the actual needs, a trial and error method may be used to determine the number of neurons in the hidden layer.
The specific mode is roughly as follows:
firstly, setting a smaller neuron number for a hidden layer to carry out network training;
then, gradually increasing the neuron number of the hidden layer, and performing network training by using the same training data;
and finally, selecting the neuron data corresponding to the minimum training error as the neuron number of the hidden layer.
And a substep S104 of determining the initial network structure according to the neuron number of the hidden layer.
It should be understood that the above is only one implementation manner for determining the initial network structure of the geological monitoring BP neural network to be optimized, and the technical solution of the present invention is not limited in any way, and in a specific application, a person skilled in the art may set the initial network structure as needed, and the present invention is not limited thereto.
It is not difficult to find out through the above description that, according to the optimization method of the geological monitoring BP neural network provided by this embodiment, when the initial network structure of the geological monitoring BP neural network to be optimized is determined, the value range of the neuron number of the hidden layer is determined according to the neuron numbers of the input layer and the output layer, and then according to the minimum mean square error of each value in the value range, the neuron number corresponding to the minimum mean square error is finally used as the neuron number of the hidden layer, so that the finally determined initial network structure of the geological monitoring BP neural network to be optimized can be closer to the actual measurement requirement, and thus the geological information of the soil to be monitored can be reasonably predicted by the geological monitoring BP neural network after being finally optimized, and the accuracy of the prediction result is greatly ensured.
Furthermore, an embodiment of the present invention further provides a computer-readable storage medium, where an optimization program of the geological monitoring BP neural network is stored on the computer-readable storage medium, and when executed by a processor, the optimization program of the geological monitoring BP neural network implements the steps of the optimization method of the geological monitoring BP neural network as described above.
Referring to fig. 4, fig. 4 is a structural block diagram of the optimization apparatus for geological monitoring BP neural network according to the first embodiment of the present invention.
As shown in fig. 4, the optimization apparatus for geological monitoring BP neural network according to the embodiment of the present invention includes: a first determining module 4001, an optimizing module 4002, an obtaining module 4003, a training module 4004, and a second determining module 4005.
The first determining module 4001 is configured to determine an initial network structure of a geological monitoring BP neural network to be optimized.
And the optimization module 4002 is configured to perform global evolution optimization on the initial weight and the initial threshold of the initial network structure by using a hill-climbing algorithm, a genetic algorithm, and a simulated annealing algorithm to obtain an optimized weight and an optimized threshold.
An obtaining module 4003 is configured to obtain training data.
It should be understood that, since the optimization device of the geological monitoring BP neural network described in this embodiment is mainly used for monitoring the geological information of the soil to be monitored, the training data needs to include longitude information, latitude information, elevation information, and belonging functional area information of the land with known monitoring results.
Correspondingly, the training module 4004 is configured to perform iterative training on the initial network structure according to the longitude information, the latitude information, the altitude information, and the functional area information, and modify the optimal weight and the optimal threshold in a training process until the iteration count reaches a preset maximum iteration count or an iteration stop condition is met, and stop training.
A second determining module 4005, configured to use the network structure when the training is stopped as the optimized network structure of the geological monitoring BP neural network to be optimized.
It should be understood that each module referred to in this embodiment is a logical module, and in practical applications, one logical unit may be one physical unit, may be a part of one physical unit, and may also be implemented by a combination of multiple physical units. In addition, in order to highlight the innovative part of the present invention, a unit which is not so closely related to solve the technical problem proposed by the present invention is not introduced in the present embodiment, but it does not indicate that there is no other unit in the present embodiment.
In addition, in order to facilitate understanding of a specific processing flow of each functional module in a practical application of the optimization apparatus for geological monitoring BP neural network provided in this embodiment, the following specifically describes processing of the optimization module 4002, the acquisition module 4003, and the training module 4004.
Specifically, the optimization module 4002 performs global evolution optimization on the initial weight and the initial threshold of the initial network structure by using a hill-climbing algorithm, a genetic algorithm, and a simulated annealing algorithm to obtain operations of optimizing the weight and the threshold, and the implementation flow in specific applications is substantially as follows:
(1) And carrying out real number coding on each neuron in the initial network structure by adopting the genetic algorithm to obtain a population data set corresponding to the initial network structure.
(2) And performing population division on the population data set by adopting the hill climbing algorithm to obtain a large population and a small population.
(3) And local optimization is carried out on the small population by adopting the hill-climbing algorithm and the genetic algorithm, and the found local optimal weight and local optimal threshold are assigned to the large population.
Specifically, the local optimization operation for the small population by using the hill-climbing algorithm and the genetic algorithm is roughly as follows:
firstly, setting a first predicted value and a first expected value for the small population;
then, calculating a first fitness value of each neuron in the small population according to the first predicted value, the first expected value and a first fitness function;
then, repeatedly performing selection, crossing and mutation genetic operations on the first fitness value of each neuron in the small population by adopting the genetic algorithm;
and finally, performing local optimization on the small population after each genetic operation by adopting the hill climbing algorithm until the local optimal weight and the local optimal threshold are obtained.
And carrying out global optimization on the large population by adopting the genetic algorithm and the simulated annealing algorithm, and taking the searched global optimal weight and global optimal threshold as the optimal weight and optimal threshold of the initial network structure.
Specifically, the above operation of global optimization of the large population by using the genetic algorithm and the simulated annealing algorithm is roughly as follows:
firstly, setting a second predicted value and a second expected value for the large population;
then, calculating a second fitness value of each neuron in the large population according to the second predicted value, the second expected value and a second fitness function;
then, repeatedly performing selection, crossing and mutation genetic operations on the second fitness value of each neuron in the large population by adopting the genetic algorithm;
and finally, carrying out global optimization on the large population after each genetic operation by adopting a simulated annealing algorithm until the global optimal weight and the global optimal threshold are obtained.
It should be understood that the above is only a specific implementation manner for performing global optimization on the initial weight and the initial threshold of the initial network structure by using a hill-climbing algorithm, a genetic algorithm, and a simulated annealing algorithm to obtain the optimized weight and the optimized threshold, and the technical scheme of the present invention is not limited in any way, and in a specific application, a person skilled in the art may set the optimal weight and the optimized threshold as needed, which is not limited by the present invention.
In addition, regarding the operation of acquiring the training data by the acquiring module 4003, the implementation flow in a specific application is roughly as follows:
firstly, acquiring sample data;
and then, carrying out maximum and minimum normalization processing on the sample data to further obtain the training data.
It should be understood that the above is only a specific implementation manner for acquiring the training data, and the technical solution of the present invention is not limited in any way, and in a specific application, a person skilled in the art may set the implementation manner as needed, and the present invention is not limited thereto.
In addition, regarding the operation of the training module 4004 performing iterative training on the initial network structure according to the longitude information, the latitude information, the altitude information, and the functional area information, the implementation flow in a specific application is substantially as follows:
and inputting the longitude information, the latitude information, the altitude information and the functional area information as input parameters into the initial network structure, and repeatedly carrying out forward propagation and backward propagation training operations on the initial network structure.
It should be understood that the above is only a specific implementation manner of performing iterative training on the initial network structure, and the technical solution of the present invention is not limited in any way, and in a specific application, a person skilled in the art may set the implementation manner as needed, and the present invention is not limited to this.
As can be easily found from the above description, when the geological monitoring BP neural network to be optimized is optimized, the optimization device for the geological monitoring BP neural network provided in this embodiment combines the hill-climbing algorithm, the genetic algorithm, and the simulated annealing algorithm with the BP neural network to realize local and global search, so that compared with the BP neural network optimized by using the genetic algorithm alone, the optimized BP neural network can effectively improve global convergence and avoid falling into a local extremum.
It should be noted that the above-described work flows are only exemplary, and do not limit the scope of the present invention, and in practical applications, a person skilled in the art may select some or all of them to achieve the purpose of the solution of the embodiment according to actual needs, and the present invention is not limited herein.
In addition, the technical details that are not described in detail in this embodiment may be referred to the optimization method of the geological monitoring BP neural network provided in any embodiment of the present invention, and are not described herein again.
Based on the first embodiment of the optimization device of the geological monitoring BP neural network, the second embodiment of the optimization device of the geological monitoring BP neural network is provided.
It should be noted that, in practical applications, the accuracy of predicting the geological information of the soil to be monitored by the geological monitoring BP neural network mainly depends on the number of neurons in the hidden layer of the geological monitoring BP neural network, so that the determined initial network structure can be fit to the reality as much as possible, based on the first embodiment, the embodiment provides an implementation manner for determining the initial network structure of the geological monitoring BP neural network to be optimized.
Correspondingly, the first determining module may be specifically divided into an input layer neuron number determining submodule, an output layer neuron number determining submodule, an implicit layer neuron value range determining submodule, an implicit layer neuron number determining submodule, and an initial network structure determining submodule.
Specifically, the input layer neuron number determining submodule is used for determining the neuron number of the input layer of the geological monitoring BP neural network to be optimized.
And the output layer neuron number determining submodule is used for determining the neuron number of the input layer and the neuron number of the output layer of the geological monitoring BP neural network to be optimized.
And the hidden layer neuron value range determining submodule is used for determining the value range of the neuron number of the hidden layer of the geological monitoring BP neural network to be optimized according to the neuron number of the input layer and the neuron number of the output layer.
And the hidden layer neuron number determining submodule is used for acquiring the minimum mean square error of each value in the value range, and taking the neuron number corresponding to the minimum mean square error as the neuron number of the hidden layer.
The initial network structure determining submodule is used for determining the initial network structure according to the neuron number of the hidden layer.
It should be understood that each module referred to in this embodiment is a logical module, and in practical applications, one logical unit may be one physical unit, may be a part of one physical unit, and may also be implemented by a combination of multiple physical units. In addition, in order to highlight the innovative part of the present invention, a unit which is not so closely related to solve the technical problem proposed by the present invention is not introduced in the present embodiment, but it does not indicate that there is no other unit in the present embodiment.
It is not difficult to find out through the above description that, in the optimization device of the geological monitoring BP neural network provided in this embodiment, when the initial network structure of the geological monitoring BP neural network to be optimized is determined, the value range of the neuron number of the hidden layer is determined according to the neuron numbers of the input layer and the output layer, and then according to the minimum mean square error of each value in the value range, the neuron number corresponding to the minimum mean square error is finally used as the neuron number of the hidden layer, so that the finally determined initial network structure of the geological monitoring BP neural network to be optimized can be closer to the actual measurement requirement, and thus the geological information of the soil to be monitored can be reasonably predicted by the geological monitoring BP neural network after being finally optimized, and the accuracy of the prediction result is greatly ensured.
It should be noted that the above-mentioned work flows are only illustrative and do not limit the scope of the present invention, and in practical applications, those skilled in the art may select some or all of them according to actual needs to implement the purpose of the solution of the present embodiment, and the present invention is not limited herein.
In addition, the technical details that are not described in detail in this embodiment can be referred to the optimization method of the geological monitoring BP neural network provided in any embodiment of the present invention, and are not described herein again.
Further, it is to be noted that, in this document, the terms "comprises," "comprising," or any other variation thereof, are intended to cover a non-exclusive inclusion, such that a process, method, article, or system that comprises a list of elements does not include only those elements but may include other elements not expressly listed or inherent to such process, method, article, or system. Without further limitation, an element defined by the phrases "comprising a," "8230," "8230," or "comprising" does not exclude the presence of other like elements in a process, method, article, or system comprising the element.
The above-mentioned serial numbers of the embodiments of the present invention are merely for description and do not represent the merits of the embodiments.
Through the above description of the embodiments, those skilled in the art will clearly understand that the method of the above embodiments can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware, but in many cases, the former is a better implementation manner. Based on such understanding, the technical solution of the present invention or portions thereof that contribute to the prior art may be embodied in the form of a software product, where the computer software product is stored in a storage medium (e.g. Read Only Memory (ROM)/RAM, magnetic disk, optical disk), and includes several instructions for enabling a terminal device (e.g. a mobile phone, a computer, a server, or a network device) to execute the method according to the embodiments of the present invention.
The above description is only a preferred embodiment of the present invention, and is not intended to limit the scope of the present invention, and all equivalent structures or equivalent processes performed by the present invention or directly or indirectly applied to other related technical fields are also included in the scope of the present invention.
Claims (10)
1. An optimization method for a geological monitoring BP neural network is characterized by comprising the following steps:
determining an initial network structure of a geological monitoring BP neural network to be optimized;
global optimization is carried out on the initial weight and the initial threshold of the initial network structure by adopting a hill climbing algorithm, a genetic algorithm and a simulated annealing algorithm to obtain an optimized weight and an optimized threshold;
acquiring training data, wherein the training data comprises longitude information, latitude information, elevation information and affiliated functional area information of the land with known monitoring results;
performing iterative training on the initial network structure according to the longitude information, the latitude information, the altitude information and the functional area information, and modifying an optimal weight and an optimal threshold in the training process until the iteration times reach a preset maximum iteration time or an iteration stopping condition is met;
and taking the network structure when the training is stopped as the optimized network structure of the geological monitoring BP neural network to be optimized.
2. The method of claim 1, wherein the step of determining an initial network structure of a geological monitoring (BP) neural network to be optimized comprises:
determining the neuron number of an input layer and the neuron number of an output layer of the geological monitoring BP neural network to be optimized;
determining the value range of the neuron number of the hidden layer of the geological monitoring BP neural network to be optimized according to the neuron number of the input layer and the neuron number of the output layer;
acquiring the minimum mean square error of each value in the value range, and taking the neuron number corresponding to the minimum mean square error as the neuron number of the hidden layer;
and determining the initial network structure according to the neuron number of the hidden layer.
3. The method of claim 2, wherein the step of using a hill-climbing algorithm, a genetic algorithm, and a simulated annealing algorithm to globally optimize the initial weights and initial thresholds of the initial network structure to obtain the optimized weights and optimized thresholds comprises:
performing real number coding on each neuron in the initial network structure by adopting the genetic algorithm to obtain a population data set corresponding to the initial network structure;
performing population division on the population data set by adopting the hill climbing algorithm to obtain a large population and a small population;
local optimization is carried out on the small population by adopting the hill climbing algorithm and the genetic algorithm, and the found local optimal weight and the local optimal threshold are assigned to the large population;
and performing global optimization on the large population by adopting the genetic algorithm and the simulated annealing algorithm, and taking the searched global optimal weight and global optimal threshold as the optimal weight and optimal threshold of the initial network structure.
4. The method of claim 3, wherein said step of using said hill-climbing algorithm and said genetic algorithm to locally optimize said small population comprises:
setting a first predicted value and a first expected value for the small population;
calculating a first fitness value of each neuron in the small population according to the first predicted value, the first expected value and a first fitness function;
repeatedly performing selection, crossing and mutation genetic operations on the first fitness value of each neuron in the small population by adopting the genetic algorithm;
and local optimization is carried out on the small population after each genetic operation by adopting the hill climbing algorithm until the local optimal weight and the local optimal threshold are obtained.
5. The method of claim 3, wherein said step of using said genetic algorithm and said simulated annealing algorithm to globally optimize said large population comprises:
setting a second predicted value and a second expected value for the large population;
calculating a second fitness value of each neuron in the large population according to the second predicted value, the second expected value and a second fitness function;
repeatedly performing selection, crossing and mutation genetic operations on the second fitness value of each neuron in the large population by adopting the genetic algorithm;
and carrying out global optimization on the large population after each genetic operation by adopting a simulated annealing algorithm until the global optimal weight and the global optimal threshold are obtained.
6. The method of any of claims 1 to 5, wherein the step of obtaining training data comprises:
and acquiring sample data, and performing maximum and minimum normalization processing on the sample data to obtain the training data.
7. The method of any of claims 1 to 5, wherein the step of iteratively training the initial network structure based on the longitude information, the latitude information, the altitude information, and the functional area information comprises:
and inputting the longitude information, the latitude information, the altitude information and the functional area information as input parameters into the initial network structure, and repeatedly carrying out forward propagation and backward propagation training operations on the initial network structure.
8. An optimization apparatus for geological monitoring BP neural network, characterized in that the apparatus comprises:
the system comprises a first determination module, a second determination module and a third determination module, wherein the first determination module is used for determining an initial network structure of a geological monitoring BP neural network to be optimized;
the optimization module is used for carrying out global evolution optimization on the initial weight and the initial threshold of the initial network structure by adopting a hill-climbing algorithm, a genetic algorithm and a simulated annealing algorithm to obtain an optimized weight and an optimized threshold;
the acquisition module is used for acquiring training data, wherein the training data comprises longitude information, latitude information and elevation information of land with known monitoring results and function area information of the land;
the training module is used for carrying out iterative training on the initial network structure according to the longitude information, the latitude information, the altitude information and the functional area information, and modifying an optimal weight and an optimal threshold value in the training process until the iteration number reaches a preset maximum iteration number or an iteration stopping condition is met;
and the second determining module is used for taking the network structure when the training is stopped as the optimized network structure of the geological monitoring BP neural network to be optimized.
9. An optimization device for geological monitoring (BP) neural network, characterized in that the device comprises: a memory, a processor and an optimizer of a geological monitoring BP neural network stored on the memory and operable on the processor, the optimizer of the geological monitoring BP neural network being configured to implement the steps of the optimization method of the geological monitoring BP neural network according to any of claims 1 to 7.
10. A computer readable storage medium, characterized in that the computer readable storage medium has stored thereon an optimization program of a geological monitoring BP neural network, which when executed by a processor, implements the steps of the optimization method of the geological monitoring BP neural network as claimed in any one of claims 1 to 7.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910525396.4A CN110232444B (en) | 2019-06-17 | 2019-06-17 | Geological monitoring BP neural network optimization method, device, equipment and storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN201910525396.4A CN110232444B (en) | 2019-06-17 | 2019-06-17 | Geological monitoring BP neural network optimization method, device, equipment and storage medium |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN110232444A CN110232444A (en) | 2019-09-13 |
| CN110232444B true CN110232444B (en) | 2023-03-28 |
Family
ID=67859469
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN201910525396.4A Active CN110232444B (en) | 2019-06-17 | 2019-06-17 | Geological monitoring BP neural network optimization method, device, equipment and storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN110232444B (en) |
Families Citing this family (5)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN111324989B (en) * | 2020-03-19 | 2024-01-30 | 重庆大学 | Gear contact fatigue life prediction method based on GA-BP neural network |
| CN111336935A (en) * | 2020-03-31 | 2020-06-26 | 上海市计量测试技术研究院 | Nano-film parameter inversion calculation method based on improved hybrid optimization algorithm |
| CN112700060B (en) * | 2021-01-08 | 2023-06-13 | 佳源科技股份有限公司 | Station terminal load prediction method and prediction device |
| CN112800673A (en) * | 2021-01-27 | 2021-05-14 | 昆明理工大学 | Method for predicting blasting block degree based on SA-GA-BP algorithm |
| CN116502286B (en) * | 2023-05-24 | 2023-11-17 | 中国标准化研究院 | Standard information service method and system based on edge calculation |
Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104866904A (en) * | 2015-06-16 | 2015-08-26 | 中电科软件信息服务有限公司 | Parallelization method of BP neural network optimized by genetic algorithm based on spark |
| CN108876054A (en) * | 2018-07-06 | 2018-11-23 | 国网河南省电力公司郑州供电公司 | Short-Term Load Forecasting Method based on improved adaptive GA-IAGA optimization extreme learning machine |
Family Cites Families (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| EP3282781A4 (en) * | 2015-05-22 | 2018-05-16 | Huawei Technologies Co., Ltd. | Method and device for improving positioning performance of artificial neural network |
-
2019
- 2019-06-17 CN CN201910525396.4A patent/CN110232444B/en active Active
Patent Citations (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN104866904A (en) * | 2015-06-16 | 2015-08-26 | 中电科软件信息服务有限公司 | Parallelization method of BP neural network optimized by genetic algorithm based on spark |
| CN108876054A (en) * | 2018-07-06 | 2018-11-23 | 国网河南省电力公司郑州供电公司 | Short-Term Load Forecasting Method based on improved adaptive GA-IAGA optimization extreme learning machine |
Also Published As
| Publication number | Publication date |
|---|---|
| CN110232444A (en) | 2019-09-13 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN110232444B (en) | Geological monitoring BP neural network optimization method, device, equipment and storage medium | |
| Goolsby et al. | Rphylopars: fast multivariate phylogenetic comparative methods for missing data and within‐species variation | |
| KR102302609B1 (en) | Neural Network Architecture Optimization | |
| Shukla | Neuro-genetic prediction of software development effort | |
| Peterman et al. | A comparison of popular approaches to optimize landscape resistance surfaces | |
| CN108021983A (en) | Neural framework search | |
| CN110766142A (en) | Model generation method and device | |
| CN111124840A (en) | Method and device for predicting alarm in business operation and maintenance and electronic equipment | |
| CN113361680B (en) | Neural network architecture searching method, device, equipment and medium | |
| JP6775469B2 (en) | OD traffic predictors, methods, and programs | |
| CN111406264A (en) | Neural architecture search | |
| Minoarivelo et al. | Detecting phylogenetic signal in mutualistic interaction networks using a Markov process model | |
| CN110705688A (en) | Neural network system, method and device for risk assessment of operation event | |
| Ding et al. | Conditional generative adversarial network model for simulating intensity measures of aftershocks | |
| Lobato et al. | Identification of an epidemiological model to simulate the COVID‐19 epidemic using robust multiobjective optimization and stochastic fractal search | |
| CN115017511A (en) | Source code vulnerability detection method and device and storage medium | |
| Czajkowski et al. | Steering the interpretability of decision trees using lasso regression-an evolutionary perspective | |
| CN110955765A (en) | Corpus construction method and apparatus of intelligent assistant, computer device and storage medium | |
| Huang et al. | Harnessing deep learning for population genetic inference | |
| KR102589525B1 (en) | Method and apparatus for providing earthquake fragility function, learning method for providing earthquake fragility function | |
| CN114338416A (en) | Space-time multi-index prediction method and device and storage medium | |
| CN114445692B (en) | Image recognition model construction method and device, computer equipment and storage medium | |
| CN115544307A (en) | Directed graph data feature extraction and expression method and system based on incidence matrix | |
| CN114625886A (en) | Entity query method and system based on knowledge graph small sample relation learning model | |
| Stork et al. | Surrogate-assisted learning of neural networks |
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 |