CN118133002A - Power plant environment-friendly index monitoring and early warning method and device based on artificial intelligence - Google Patents
Power plant environment-friendly index monitoring and early warning method and device based on artificial intelligence Download PDFInfo
- Publication number
- CN118133002A CN118133002A CN202410571507.6A CN202410571507A CN118133002A CN 118133002 A CN118133002 A CN 118133002A CN 202410571507 A CN202410571507 A CN 202410571507A CN 118133002 A CN118133002 A CN 118133002A
- Authority
- CN
- China
- Prior art keywords
- target
- monitoring
- early warning
- power plant
- sample data
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Granted
Links
- 238000012544 monitoring process Methods 0.000 title claims abstract description 117
- 238000000034 method Methods 0.000 title claims abstract description 72
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 24
- 238000012545 processing Methods 0.000 claims abstract description 78
- 230000007613 environmental effect Effects 0.000 claims abstract description 66
- 238000012549 training Methods 0.000 claims abstract description 61
- 238000004422 calculation algorithm Methods 0.000 claims abstract description 59
- 230000002159 abnormal effect Effects 0.000 claims abstract description 13
- 230000002708 enhancing effect Effects 0.000 claims abstract description 9
- 238000005457 optimization Methods 0.000 claims description 81
- 229930192334 Auxin Natural products 0.000 claims description 72
- 239000002363 auxin Substances 0.000 claims description 72
- SEOVTRFCIGRIMH-UHFFFAOYSA-N indole-3-acetic acid Chemical compound C1=CC=C2C(CC(=O)O)=CNC2=C1 SEOVTRFCIGRIMH-UHFFFAOYSA-N 0.000 claims description 72
- 238000013528 artificial neural network Methods 0.000 claims description 60
- 230000009467 reduction Effects 0.000 claims description 40
- 241000282461 Canis lupus Species 0.000 claims description 34
- 239000013598 vector Substances 0.000 claims description 32
- 238000009826 distribution Methods 0.000 claims description 25
- 238000000605 extraction Methods 0.000 claims description 21
- 230000008569 process Effects 0.000 claims description 20
- 238000013508 migration Methods 0.000 claims description 17
- 230000005012 migration Effects 0.000 claims description 17
- 230000007246 mechanism Effects 0.000 claims description 16
- 238000003860 storage Methods 0.000 claims description 10
- 238000001514 detection method Methods 0.000 claims description 4
- 230000006870 function Effects 0.000 description 23
- 238000004364 calculation method Methods 0.000 description 9
- 230000000875 corresponding effect Effects 0.000 description 9
- 230000000694 effects Effects 0.000 description 9
- 238000011156 evaluation Methods 0.000 description 8
- 241000282421 Canidae Species 0.000 description 4
- 230000003044 adaptive effect Effects 0.000 description 4
- 230000006399 behavior Effects 0.000 description 4
- 230000008901 benefit Effects 0.000 description 4
- 230000024245 cell differentiation Effects 0.000 description 4
- 238000004891 communication Methods 0.000 description 4
- 230000001276 controlling effect Effects 0.000 description 4
- 238000013461 design Methods 0.000 description 4
- 238000010586 diagram Methods 0.000 description 4
- 230000006872 improvement Effects 0.000 description 4
- 230000001105 regulatory effect Effects 0.000 description 4
- 230000009471 action Effects 0.000 description 3
- 230000004663 cell proliferation Effects 0.000 description 3
- 238000004590 computer program Methods 0.000 description 3
- 238000007405 data analysis Methods 0.000 description 3
- 238000011161 development Methods 0.000 description 3
- 230000018109 developmental process Effects 0.000 description 3
- 230000035755 proliferation Effects 0.000 description 3
- 238000005070 sampling Methods 0.000 description 3
- 230000035945 sensitivity Effects 0.000 description 3
- 238000012546 transfer Methods 0.000 description 3
- 230000008859 change Effects 0.000 description 2
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000005265 energy consumption Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000004519 manufacturing process Methods 0.000 description 2
- 238000003062 neural network model Methods 0.000 description 2
- 230000027870 phototropism Effects 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 238000012795 verification Methods 0.000 description 2
- XLYOFNOQVPJJNP-UHFFFAOYSA-N water Substances O XLYOFNOQVPJJNP-UHFFFAOYSA-N 0.000 description 2
- 101150055297 SET1 gene Proteins 0.000 description 1
- 230000006978 adaptation Effects 0.000 description 1
- 238000004458 analytical method Methods 0.000 description 1
- 230000008485 antagonism Effects 0.000 description 1
- 230000009286 beneficial effect Effects 0.000 description 1
- 230000015556 catabolic process Effects 0.000 description 1
- 238000013145 classification model Methods 0.000 description 1
- 238000011217 control strategy Methods 0.000 description 1
- 230000002596 correlated effect Effects 0.000 description 1
- 238000013480 data collection Methods 0.000 description 1
- 230000003247 decreasing effect Effects 0.000 description 1
- 238000006731 degradation reaction Methods 0.000 description 1
- 230000004069 differentiation Effects 0.000 description 1
- 238000005315 distribution function Methods 0.000 description 1
- 238000003912 environmental pollution Methods 0.000 description 1
- 238000000802 evaporation-induced self-assembly Methods 0.000 description 1
- 238000001914 filtration Methods 0.000 description 1
- 238000011478 gradient descent method Methods 0.000 description 1
- 238000005286 illumination Methods 0.000 description 1
- 238000012804 iterative process Methods 0.000 description 1
- 230000007786 learning performance Effects 0.000 description 1
- 238000007477 logistic regression Methods 0.000 description 1
- 239000011159 matrix material Substances 0.000 description 1
- 238000012986 modification Methods 0.000 description 1
- 230000004048 modification Effects 0.000 description 1
- 230000035772 mutation Effects 0.000 description 1
- 238000010606 normalization Methods 0.000 description 1
- 230000003287 optical effect Effects 0.000 description 1
- 238000003909 pattern recognition Methods 0.000 description 1
- 239000003375 plant hormone Substances 0.000 description 1
- 238000007781 pre-processing Methods 0.000 description 1
- 230000005180 public health Effects 0.000 description 1
- 238000011946 reduction process Methods 0.000 description 1
- 230000010076 replication Effects 0.000 description 1
Classifications
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y04—INFORMATION OR COMMUNICATION TECHNOLOGIES HAVING AN IMPACT ON OTHER TECHNOLOGY AREAS
- Y04S—SYSTEMS INTEGRATING TECHNOLOGIES RELATED TO POWER NETWORK OPERATION, COMMUNICATION OR INFORMATION TECHNOLOGIES FOR IMPROVING THE ELECTRICAL POWER GENERATION, TRANSMISSION, DISTRIBUTION, MANAGEMENT OR USAGE, i.e. SMART GRIDS
- Y04S10/00—Systems supporting electrical power generation, transmission or distribution
- Y04S10/50—Systems or methods supporting the power network operation or management, involving a certain degree of interaction with the load-side end user applications
Landscapes
- Management, Administration, Business Operations System, And Electronic Commerce (AREA)
Abstract
The invention provides a power plant environmental protection index monitoring and early warning method and device based on artificial intelligence, which relate to the technical field of data processing and comprise the following steps: determining the power plant environment monitoring data sent by a power plant data sensor as an original sample data set; generating an countermeasure network algorithm through the robustness enhancement, after enhancing the robustness of noise and abnormal values in an original sample data set by the generated countermeasure network, performing data expansion processing on the original sample data set, and determining a target sample data set; based on the target sample data set, model training processing is carried out on a preset power plant environmental protection index monitoring and early warning model, and a target monitoring and early warning model is determined, wherein the target monitoring and early warning model is used for carrying out index monitoring processing on the power plant environmental protection index, and a monitoring and early warning result is determined. The method can remarkably improve the accuracy of monitoring and early warning of the environmental protection indexes of the power plant.
Description
Technical Field
The invention relates to the technical field of data processing, in particular to an artificial intelligence-based power plant environmental protection index monitoring and early warning method and device.
Background
With the global improvement of environmental protection awareness and the strict execution of environmental regulations, energy production and conversion facilities such as power plants face increasingly strict environmental protection index monitoring and emission standards, and the data volume generated by power plant environmental monitoring is huge and complex, which requires a monitoring and early warning system to be capable of processing high-dimensional data, and also to have efficient data analysis and processing capability and accurate prediction and classification capability.
At present, related technologies propose that an energy consumption calculation model can be built according to operation data of power plant equipment, energy efficiency evaluation indexes are manually built according to model operation effects, so that influence rules of different factors on energy consumption are obtained, and an optimal control strategy is proposed according to actual operation conditions of a power plant.
Disclosure of Invention
Therefore, the invention aims to provide the power plant environmental protection index monitoring and early warning method and device based on artificial intelligence, which can remarkably improve the accuracy of the power plant environmental protection index monitoring and early warning.
In a first aspect, an embodiment of the present invention provides an artificial intelligence-based power plant environmental protection index monitoring and early warning method, including: determining the power plant environment monitoring data sent by a power plant data sensor as an original sample data set; generating an countermeasure network algorithm through the robustness enhancement, after enhancing the robustness of noise and abnormal values in an original sample data set by the generated countermeasure network, performing data expansion processing on the original sample data set, and determining a target sample data set; based on the target sample data set, model training processing is carried out on a preset power plant environmental protection index monitoring and early warning model, and a target monitoring and early warning model is determined, wherein the target monitoring and early warning model is used for carrying out index monitoring processing on the power plant environmental protection index, and a monitoring and early warning result is determined.
In one embodiment, based on the target sample data set, performing model training processing on a preset power plant environmental protection index monitoring and early warning model, and determining the target monitoring and early warning model, the method comprises the following steps: performing feature extraction processing on a target sample data set based on a neural network parameter optimization algorithm of auxin regulation optimization, and determining a feature set; inputting the target sample data set and the feature set into a preset classifier, and performing classification training on the preset classifier based on an improved anchor point and wolf optimization algorithm to determine a target monitoring and early warning model, wherein the improved anchor point and wolf optimization algorithm is used for optimizing the weight and bias parameters of the preset classifier so as to improve the detection performance of the target monitoring and early warning model.
In one embodiment, the step of performing feature extraction processing on the target sample data set and determining a feature set based on a neural network parameter optimization algorithm for auxin regulation optimization includes: adjusting parameters of the neural network by using a neural network parameter optimization algorithm based on auxin regulation and optimization and simulating a distribution and migration mechanism of auxin so as to perform parameter optimization adjustment processing on weights and biases of the neural network and determine a target neural network; and performing feature extraction processing on the target sample data set by using the target neural network to determine a feature set.
In one embodiment, the step of determining the target neural network by using a neural network parameter optimization algorithm based on auxin regulation optimization to adjust the neural network parameters by simulating the distribution and migration mechanism of auxin to perform parameter optimization adjustment processing on the weights and biases of the neural network comprises the following steps: simulating the distribution condition of initialized auxin in a preset parameter space, wherein each parameter vector in the preset parameter space is used for simulating the cell position in a plant body, the corresponding fitness of each parameter vector is used for simulating the auxin concentration of the cell position, the parameter vector is used for representing the weight and bias of a neural network, and the fitness is positively related to the auxin concentration; and (3) performing parameter optimization adjustment processing on the neural network parameters by simulating the migration process of auxin between cells from a high-concentration auxin region to a low-concentration auxin region, and determining the target neural network.
In one embodiment, the target sample data set and the feature set are input into a preset classifier, and before the step of classifying and training the preset classifier based on the improved anchor point gray wolf optimization algorithm to determine the target monitoring and early warning model, the method comprises the following steps: inputting the feature set into a preset feature dimension reduction model, performing data noise reduction processing on the feature set by using a three-dimensional spiral projection self-encoder in the preset feature dimension reduction model, and determining a target feature set after noise reduction.
In one embodiment, the step of inputting the feature set into a preset feature dimension reduction model, performing data noise reduction processing on the feature set by using a stereo spiral projection self-encoder in the preset feature dimension reduction model, and determining a target feature set after noise reduction includes: projecting the feature set to a stereoscopic spiral space through a stereoscopic spiral projection self-encoder, and carrying out parameter adjustment processing on the shape and the size of the spiral according to feature dynamics corresponding to each data in the stereoscopic spiral space to determine a target stereoscopic spiral space, wherein the feature dynamics comprise: radius features, helix angle features, angular rotation rate features, and height features along the helical axis; and in the target stereo spiral space, performing feature dimension reduction processing on the projected data by using a preset low-dimensional coding learning model to remove redundant information and noise, and reserving a target feature set containing key information.
In one embodiment, the step of inputting the target sample data set and the feature set into a preset classifier, performing classification training on the preset classifier based on an improved anchor point gray wolf optimization algorithm, and determining a target monitoring and early warning model includes: inputting a target sample data set and a target feature set into a preset classifier, and determining a target importance weight based on an anchor point mechanism and a dynamic adjustment strategy through an anchor point gray wolf optimization algorithm based on improvement in the preset classifier, wherein the target importance weight is the optimal weight configuration of an extreme learning machine classifier based on the anchor point gray wolf optimization algorithm; and carrying out classification training on a preset classifier according to the importance weight of the target, and determining a target monitoring and early warning model.
In a second aspect, an embodiment of the present invention further provides an artificial intelligence-based power plant environmental protection index monitoring and early warning device, where the device includes: the sample data acquisition module is used for determining the power plant environment monitoring data sent by the power plant data sensor as an original sample data set; the data expansion module is used for carrying out data expansion processing on the original sample data set after enhancing the robustness of noise and abnormal values in the original sample data set by enhancing the generated countermeasure network algorithm through the robustness enhanced generation countermeasure network algorithm, so as to determine a target sample data set; the model training module is used for carrying out model training processing on a preset power plant environmental protection index monitoring and early warning model based on the target sample data set to determine the target monitoring and early warning model, wherein the target monitoring and early warning model is used for carrying out index monitoring processing on the power plant environmental protection index to determine a monitoring and early warning result.
In a third aspect, embodiments of the present invention also provide an electronic device comprising a processor and a memory storing computer executable instructions executable by the processor, the processor executing the computer executable instructions to implement the method of any one of the first aspects.
In a fourth aspect, embodiments of the present invention also provide a computer-readable storage medium storing computer-executable instructions that, when invoked and executed by a processor, cause the processor to implement the method of any one of the first aspects.
The embodiment of the invention has the following beneficial effects:
According to the power plant environmental protection index monitoring and early warning method and device based on artificial intelligence, after power plant environmental protection index monitoring data sent by a power plant data sensor are determined to be an original sample data set, an antagonism network algorithm is generated through robustness enhancement, after the robustness of noise and abnormal values in the original sample data set is enhanced, data expansion processing is carried out on the original sample data set, a target sample data set is determined, model training processing is carried out on a preset power plant environmental protection index monitoring and early warning model based on the target sample data set, and a target monitoring and early warning model is determined.
Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and claims hereof as well as the appended drawings.
In order to make the above objects, features and advantages of the present invention more comprehensible, preferred embodiments accompanied with figures are described in detail below.
Drawings
In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings that are needed in the description of the embodiments or the prior art will be briefly described, and it is obvious that the drawings in the description below are some embodiments of the present invention, and other drawings can be obtained according to the drawings without inventive effort for a person skilled in the art.
FIG. 1 is a schematic flow chart of an artificial intelligence-based power plant environmental protection index monitoring and early warning method provided by an embodiment of the invention;
FIG. 2 is a schematic flow chart of a model training method according to an embodiment of the present invention;
FIG. 3 is a schematic structural diagram of an artificial intelligence-based power plant environmental protection index monitoring and early warning device according to an embodiment of the present invention;
fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.
Detailed Description
For the purpose of making the objects, technical solutions and advantages of the embodiments of the present invention more apparent, the technical solutions of the present invention will be clearly and completely described in conjunction with the embodiments, and it is apparent that the described embodiments are some embodiments of the present invention, but not all embodiments. All other embodiments, which can be made by those skilled in the art based on the embodiments of the invention without making any inventive effort, are intended to be within the scope of the invention.
Currently, with the global rise of environmental protection awareness and the strict execution of environmental regulations, energy production and conversion facilities such as power plants face increasingly strict environmental protection index monitoring and emission standards, including but not limited to, various aspects of exhaust emission concentration, emission rate, water quality index, temperature, humidity, noise level, etc., and their monitoring plays an important role in preventing environmental pollution, protecting public health and complying with legal regulations. In this context, conventional environmental indicator monitoring methods face various challenges including limitations in data collection, complexity of data processing and analysis, and insufficient sensitivity and early warning capability to abnormal conditions, especially in the big data age, the amount of data generated by environmental monitoring in a power plant is huge and complex, which requires that the monitoring early warning system not only be capable of processing data with high dimensionality, but also have efficient data analysis and processing capability, and accurate prediction and classification capability, and furthermore, environmental indicator monitoring is not only as simple as collecting and recording data, but also how to extract useful information from the data, and how to use the information for effective early warning and decision support, which requires that the monitoring early warning system have advanced data analysis, feature extraction, data degradation and pattern recognition capability.
However, the traditional method is often low in efficiency when processing such tasks, has limited accuracy and generalization capability, is difficult to meet the current requirements of environmental protection monitoring of power plants, and particularly, the prior art has limited collected data volume and diversity, and particularly, less acquired data is obtained under specific environments or special cases, which can influence the generalization capability and accuracy of the model; the traditional method has poor effect in processing the power plant monitoring data with abnormal values and noise, and can cause non-ideal model training effect and influence the final monitoring and early warning accuracy; the prior art mainly relies on traditional gradient descent or similar algorithms for parameter optimization of the neural network, and the methods are easy to fall into a local optimal solution when processing complex and high-dimensional data; the traditional feature extraction and dimension reduction method does not fully consider the inherent structure and complexity of the data, so that important information is lost in the dimension reduction process, and the performance of a subsequent classifier is affected; the existing classification method has limitations on accuracy and generalization capability, and particularly when facing complex environmental protection index data, different environmental protection risk grades cannot be accurately distinguished.
Referring to fig. 1, a flow chart of an artificial intelligence-based power plant environmental protection index monitoring and early warning method mainly includes the following steps S102 to S106:
Step S102, determining power plant environment monitoring data sent by a power plant data sensor as an original sample data set, wherein the original sample data set is derived from various data acquisition sensors of a power plant, the collected data are in a structured vector data format, the data are preprocessed and stored in a central data warehouse, the original sample data set is used for training and verifying an artificial intelligent model, and in one embodiment, the collected data comprise the following attributes: : emission concentration,/> : Discharge rate,/>: Water quality index,/>: Temperature,/>: Humidity,/>: Wind speed,/>: Wind direction,/>: Atmospheric pressure,/>: Noise level,/>: The running state of the device, in which this embodiment is only for illustrating one data format and kind of the present invention, in practical application, the attributes of the data are usually more than 10 attributes, and the number of the attributes of the data may reach tens or hundreds. Further, the collected data can be marked in a manual marking mode, wherein marking categories comprise normal emission, out-of-standard emission, equipment failure and abnormal weather influence 4 categories.
Step S104, after the robustness of noise and abnormal values in the original sample data set is enhanced through a robustness enhanced generation countermeasure network algorithm, data expansion processing is carried out on the original sample data set, and a target sample data set is determined.
Step S106, performing model training processing on a preset power plant environmental protection index monitoring and early warning model based on a target sample data set, and determining a target monitoring and early warning model, wherein the target monitoring and early warning model is used for performing index monitoring processing on the power plant environmental protection index, and determining a monitoring and early warning result.
The power plant environmental protection index monitoring and early warning method based on the artificial intelligence provided by the embodiment of the invention can obviously improve the accuracy of the power plant environmental protection index monitoring and early warning.
The embodiment of the invention also provides an algorithm training mode for generating the countermeasure network algorithm based on the robustness enhancement, which comprises the following steps (1) to (6):
(1) Initializing: initializing parameters of a generator and a discriminator, and setting the parameters of the generator and the discriminator as respectively And/>. The initialization process can be expressed as:
Initializing generator parameters:
initializing parameters of a discriminator:
Wherein, The representation obeys a specific distribution,/>Parameter initial value of generator,/>Representing the initial value of the parameters of the arbiter,/>Representing a normal distribution with a mean value of0 and standard deviation of the identity matrix, in one embodiment, the network parameters of the generator and the arbiter/>And/>The initialization adopts normal distribution with the mean value of 0 and the standard deviation of 0.02.
(2) Robustness constraints: preprocessing input data, including data normalization and noise filtering, to enhance model robustness to outliers and noise, in particularFor inputting data vector,/>For the data vector subjected to the robustness design processing, the processing function is/>The calculation mode of the data vector after the robustness design processing can be expressed as follows:
Wherein, And/>Respectively/>Mean and standard deviation of/(Is added with a minute noise to enhance the robustness, and in one embodiment, the minute noise level is set to 1% of the standard deviation of the original data.
Further, the method comprises the steps of,And/>The calculation mode of (2) is as follows:
Wherein, Is the number of samples,/>Is/>Data of individual samples.
(3) Generating countermeasure training: in each iteration, the generator attempts to generate data that is as similar as possible to the real plant monitoring data, while the arbiter attempts to distinguish between the generated data and the real data. By this countermeasure process, the parameters of the generator and of the arbiter are continuously adjusted, in particular, in the firstIn the next iteration, generator/>Generate data/>The manner of (a) can be expressed as:
Wherein, Is from a certain distribution/>Random noise vector obtained by middle sampling,/>For the data generated by the generator at the t-th iteration,/>The expression parameter is/>Is a generator function of (a).
Further, a discriminatorThe goal of (a) is to distinguish true data/>And generating data/>Its loss function/>The definition is as follows:
Wherein, Representing expectations,/>And/>Respectively represent the distribution obeyed by the input of the discriminator and the input of the generator,/>Is the dynamic weight of the arbiter at the t-th iteration.
Further, the loss function of the generatorThe definition is as follows:
Wherein, The expression parameter is/>Is a discriminant function of/>Is the dynamic weight of the generator at the t-th iteration.
Further, in generating the contrast training, expectations in the loss functionIs obtained by sampling a sample set, for/>By randomly extracting a batch of samples from a real datasetAnd the number of samples extracted is m, the average value is calculated to approximate this expected value, which can be expressed as:
And, for the following By distributing/>, from random noiseThe mid-sampling results in a set of noise/>Input into the generator to generate a permit a leave samples, and then calculate the logarithm of the average arbiter output of these false samples, which can be expressed as:
Further, dynamic weights of the arbiter and generator And/>Is based on the following strategy:
For the arbiter weights :/>
For generator weights:/>
Wherein SigIs a sigmoid function used for adjusting the weight to be in the range of (0, 1), ensures the dynamic adjustment and the rationalization of the weight of each part of the loss function,/>And/>Is a super-parameter for controlling the sensitivity of weight adjustment, determines the speed and amplitude of weight adjustment,/>Indicating that the arbiter is at the/>The accuracy in the secondary iteration is used for evaluating the performance of the discriminator, and the accuracy is obtained by a preset logistic regression classifier,/>The representation generator is at the/>The ratio of the generated data misjudged by the discriminant to the real data in the multiple iterations is used to evaluate the performance of the generator,/>Is a target threshold for generator performance, for determining whether the generator performance meets an intended target, in one particular embodiment, is setAnd/>To control the sensitivity of the weight adjustment while setting a target threshold for generator performanceI.e. it is expected that at least 80% of the generated data is misjudged by the arbiter as real data.
(4) Non-convex optimization: in the process of countermeasure training, adopting a non-convex optimization strategy to adjust the parameter update of the generator and the discriminator so as to avoid sinking into suboptimal solutions and ensure the quality of generated data, specifically adopting a non-convex optimization technology to adjust the parameters, and the parameter update mode of the discriminator and the generator can be expressed as follows:
updating parameters of a discriminator:
Updating generator parameters:
Wherein, For the pre-update arbiter parameters,/>For updated arbiter parameters,/>To update the generator parameters before-For updated generator parameters,/>And/>Learning rates of the arbiter and the generator, respectively.
Further, gradientCan be further developed as:
And, for the generator parameters Gradient/>Can be unfolded as follows:
(5) Loop iteration and enhancement: the generating countermeasure training step is repeatedly performed, and after each iteration, the strategy of the generator is adjusted according to the feedback of the arbiter, so as to enhance the diversity and the authenticity of the generated data until a termination condition is met, and in one embodiment, the termination condition is that a preset iteration number is reached.
(6) Model evaluation and adjustment: using independent verification data set to evaluate quality and diversity of generated data, adjusting model parameters and training strategies according to evaluation results to optimize generation effect, and setting evaluation indexesAnd adjusting the learning rate and other super parameters according to the evaluation result to optimize the generation effect.
Referring to the schematic flow chart of a model training method shown in fig. 2, the embodiment of the invention further provides an implementation manner of model training, and specifically refers to the following steps S202 to S206:
Step S202, optimizing parameters of the neural network by simulating the regulation and control action of auxin in the growth and development of plants so as to train a feature extraction model and perform feature extraction processing. The neural network parameter optimization method based on auxin regulation optimization is inspired in the nature, auxin is used as a plant hormone and is responsible for regulating and controlling the elongation, division, phototropism and other behaviors of plant cells, an auxin regulation optimization algorithm refers to the regulation and control actions of auxin in the growth and development of plants, parameters (weight w and bias b) of a neural network are regulated and optimized through simulating the distribution and migration mechanism of auxin, so that efficient and accurate learning performance is realized, when the neural network parameter optimization algorithm based on auxin regulation optimization is used for carrying out feature extraction processing on a target sample data set, when a feature set is determined, the neural network parameter optimization algorithm based on auxin regulation optimization can be utilized, the neural network parameter is regulated through simulating the distribution and migration mechanism of auxin, so that the parameter optimization adjustment processing is carried out on the weight and bias of the neural network, the target neural network is determined, and the feature extraction processing is carried out on the target sample data set by utilizing the target neural network model, so that the feature set is determined.
In one embodiment, the distribution condition of the auxin is firstly simulated in a preset parameter space, then parameter optimization adjustment processing is carried out on the parameters of the neural network to determine the target neural network by simulating the migration process of the auxin from a high-concentration auxin region to a low-concentration auxin region among cells, wherein each parameter vector in the preset parameter space is used for simulating the cell position in a plant body, the corresponding fitness of each parameter vector is used for simulating the auxin concentration of the cell position, the parameter vector is used for representing the weight and bias of the neural network, and the fitness is positively correlated with the auxin concentration. Since the auxin regulation and control optimization algorithm does not depend on the traditional gradient descent algorithm, parameters of the neural network are regulated by simulating the distribution and transportation mechanism of auxin in a plant body, more effective searching is facilitated in a parameter space, particularly in a complex and high-dimensional parameter space, the problem of sinking into a local optimal solution can be effectively avoided, global searching capacity is improved, and specifically, the step of training the neural network characteristic extraction model (namely, a target neural network) based on auxin regulation and control optimization comprises the following steps: initializing auxin distribution, assessing fitness, auxin migration, cell proliferation and differentiation, phototropic modulation, fitness modulation and iterative optimization.
In performing the step of initializing auxin distribution, it is necessary to simulate the distribution of initializing auxin in a parameter space, wherein each parameter vector corresponds to the position of one cell in the plant body, and the auxin concentration thereof represents the fitness of the parameter vector. In one embodiment, a parameter vector is providedRepresenting weights and biases in the neural network. In the initialization phase, for each parameter/>Assigning an initial auxin concentration/>In this embodiment, the implementation by a normal distribution function can be expressed as:
Wherein, And/>The mean and variance of the distribution, respectively. In one embodiment,/>Is set to 0,/>Is set to 1.
In performing the step of evaluating fitness, it is necessary to evaluate the performance of each cell based on the auxin concentration (i.e., fitness of the parameter vector), and cells with high fitness contain a higher concentration of auxin. Specifically, fitness functionFor evaluating the performance of a parameter vector, associated with a loss function of a neural network, set/>As a loss function, fitness can be expressed as:
Wherein, Is the cross entropy loss of the neural network.
In the step of transferring auxin, it is necessary to simulate the transfer process of auxin between cells, that is, the transfer of auxin from a region with high concentration to a region with low concentration, and to promote the adjustment of parameters in a direction of better solution, and the transfer of auxin simulates the flow of auxin from a region with high concentration to a region with low concentration. Specifically, it is provided withFor parameter/>To/>The auxin migration amount of (2)/>The calculation of (2) can be expressed as:
Wherein, Is a parameter/>And/>Distance in parameter space,/>And/>Is a hyper-parameter controlling the migration velocity and distance effects. /(I)Is a super-parameter for adjusting the influence intensity of environmental feedback,/>Is an environmental feedback signal. In one embodiment,/>Is set to 0.1,/>Is set to 2.
Further, parametersAnd/>Distance in parameter space/>The calculation of (2) can be expressed as:
Wherein, And/>Parameter vector/>, respectivelyAnd/>/>The elements.
Further, an environmental feedback signalIs at the/>The rate of change defined as the loss function in each iteration is used to measure the fluctuation of the network performance and can be expressed as:
Wherein, And/>Respectively represent the/>And/>Loss function value after several iterations. /(I)The value of (2) reflects the rate of improvement in network performance, with positive values indicating increased performance and negative values indicating decreased performance.
In the step of cell proliferation and differentiation, it is necessary to simulate proliferation and differentiation of plant cells under the influence of auxin, i.e., to replicate a well behaved parameter vector in a parameter space and introduce a minute variation to increase the diversity of searches, and proliferation and differentiation of cells (i.e., parameter vector) is achieved by replication and minute variation, specifically, provided thatFor the copied parameter vector, the calculation mode can be expressed as follows:
Wherein, Is a super-parameter for controlling variation intensity,/>Is the variance of the normal distribution used in the mutation process,/>Is the scale factor of the t-th iteration. In one embodiment,/>Is set to 0.05,/>Is set to 0.01.
Further, the variantThe variance, which is resolved into each parameter dimension, can be expressed as:
Wherein each parameter is Variation of/>Control from 0 as average,/>Is a randomly extracted value in the normal distribution of variances.
Further, the scale factor of the t-th iterationRepresents the/>Exploration scale of secondary iteration,/>Based on training progress and performance index dynamic changes, can be expressed as:
Wherein, Is the maximum scale factor for initial large scale exploration; /(I)Is a scale attenuation coefficient, and controls the attenuation speed of the scale factor along with time; /(I)Is a predetermined maximum number of training iterations for normalizing the training schedule.
In the step of phototropic adjustment, the phototropic behavior of the plant needs to be used as a reference, and the growth or adjustment of the parameter vector towards the region with higher fitness gradient (like illumination intensity) is simulated, and the phototropic adjustment simulates the growth of the parameter towards the region with higher fitness gradient. Specifically, it is provided withFor the adaptation function to the gradient of the parameter vector, the update of the parameter can be expressed as:
Wherein, For updated parameter vector,/>Is the learning rate, controls the step length of parameter updating. In one embodiment,/>Is set to 0.01.
Further, gradientCalculation by the chain law can be expressed as:
Wherein, Representing a loss function/>With respect to parameter vector/>Is a gradient of (a).
In the step of adaptively adjusting, parameters such as an auxin migration rate, a proliferation rate and the like need to be dynamically adjusted according to the progress of the optimization process so as to adapt to different stages in the optimization process. Adaptive tuning involves dynamically changing super-parameters, e.g.、/>And/>To accommodate the optimization process. In one embodiment, let/>For the current iteration number, the adjustment mode of the super parameter can be expressed as:
Wherein, Is the initial learning rate,/>For the learning rate of the t-th iteration,/>Is the decay rate. In one embodiment,/>Is set to 0.1,/>Set to 0.001.
Further, the attenuation ratioThe calculation of (2) can be expressed as:
Wherein, Is the total iteration number,/>Is that the learning rate is at/>A target value for the moment.
The above steps of assessing fitness, auxin migration, cell proliferation and differentiation and phototropism adjustment are repeated until a preset number of iterations is met, wherein the number of iterations may be preset to 500.
And S204, inputting the data after feature extraction into a feature dimension reduction model, and training the feature dimension reduction model. In one embodiment, unlike conventional self-encoders, the stereoscopic spiral projection self-encoder provided by the invention adopts a stereoscopic spiral projection mechanism in the encoding process to more effectively capture and represent complex structures in high-dimensional data, wherein the stereoscopic spiral projection self-encoder is characterized in that the encoder part not only learns low-dimensional representation of the data, but also projects the data into a stereoscopic spiral space with exquisite design, thereby enhancing the capture capability of a model on the internal structure of the data, so that the stereoscopic spiral projection self-encoder can retain important information of more original data when performing characteristic dimension reduction, and simultaneously remove redundancy and noise.
In one embodiment, the feature set may be input into a preset feature dimension reduction model, and the data noise reduction processing is performed on the feature set by using a stereo spiral projection self-encoder in the preset feature dimension reduction model, so as to determine a target feature set after noise reduction: projecting a feature set to a stereo spiral space through a stereo spiral projection self-encoder, carrying out parameter adjustment processing on the shape and the size of a spiral according to feature dynamics corresponding to each data in the stereo spiral space, determining a target stereo spiral space, carrying out feature dimension reduction processing on the projected data in the target stereo spiral space by using a preset low-dimensional coding learning model so as to remove redundant information and noise, and reserving a target feature set containing key information, wherein the feature dynamics comprise: radius features, helix angle features, angular rotation rate features, and height features along the helical axis, specifically, the step of training the feature dimension reduction model includes: stereo spiral projection, low-dimensional coding learning, decoding and reconstruction, loss function optimization, feedback and iteration.
Firstly, adopting a random initialization mode to randomly initialize network parameters of an encoder and a decoder of a self-encoder of stereoscopic spiral projection, and then performing stereoscopic spiral projection: the encoder projects the input power plant monitoring data after feature extraction into a three-dimensional spiral space, the three-dimensional spiral space is realized through a specially designed spiral function, the function can dynamically adjust the shape and the size of a spiral according to the characteristics of the data, and specifically, the power plant environment-friendly monitoring data after feature extraction is set asWherein subscripts/>Representing feature vectors, stereo spiral projection function/>Will/>Projection into a stereohelical space can be expressed as:
Wherein, Is the spiral radius, defined as:
Wherein, Is L1 norm sign,/>And/>Is a parameter for adjusting the radius. In one embodiment, set/>,/>。
Further, the method comprises the steps of,Is the helix angle, defined as:
Wherein, And/>Adjusting the rate of angular rotation, in one embodiment, set/>,。
Further, the method comprises the steps of,The height along the helical axis is defined as:
Wherein, And/>Is a parameter that adjusts for height variations, in one embodiment, set/>,。
Further, the method comprises the steps of,Representation/>The L1 norm of (2), namely:
/>
Wherein, Is vector/>Dimension,/>Representation/>Summation of elements.
Further, the method comprises the steps of,Representation/>The variance of (2), namely:
Wherein, Is/>Is equal to (1).
In the low-dimensional code learning step, the projected data is further compressed into a lower-dimensional space, and the projected data is realized through a traditional neural network layer and is matched with stereoscopic spiral projection, so that the key structure of the data is ensured to be reserved. In particular, an encoderData after receiving stereo helical projection/>And outputs a low-dimensional code/>The manner of (a) can be expressed as:
Wherein, Is an encoder parameter including network weights and offsets.
In performing the decoding and reconstruction steps, it is necessary for the decoder portion to receive the low-dimensional code and to attempt to reconstruct the original data. The goal of this process is to minimize the difference between the original data and the reconstructed data, in one embodiment, the metric used is the mean square error, then the decoderFrom low dimensional encoding/>Reconstruction data/>The manner of (a) can be expressed as:
Wherein, Is a decoder parameter.
In performing the step of loss function optimization, since the training objective of the overall network is to minimize the loss function, the function takes into account the accuracy of the data reconstruction and the compactness of the encoding. Integral loss functionIncluding reconstruction losses and regularization terms to ensure accuracy of data reconstruction and compactness of encoding, can be expressed as:
Wherein, Is the weight of regularization term used to control model complexity; /(I)Is the L2 norm symbol. In one embodiment,/>。
When the steps of feedback and iteration are carried out, network parameters are required to be updated through a back propagation algorithm, and in the training process, the learning rate and other super parameters are adjusted according to the performance on the verification set so as to optimize the performance of the model. Updating by gradient descent methodAnd/>Can be expressed as:
For encoder parameters:
For decoder parameters:
Wherein, Is learning rate,/>To adjust the pre-encoder parameters,/>In order to adjust the parameters of the encoder,For adjusting the pre-decoder parameters,/>For adjusted decoder parameters,/>Is a partial guide symbol.
Finally, model convergence is carried out, and the training steps are repeatedly executed until the preset maximum iteration times are reached, wherein the preset maximum iteration times can be 1000 times.
And S206, inputting the dimensionality reduced data into a classifier for training the classifier, and determining a target monitoring early warning model. Inputting the target sample data set and the characteristic set into a preset classifier, carrying out classification training on the preset classifier based on an improved anchor point gray wolf optimization algorithm, and determining a target monitoring early warning model, wherein the improved anchor point gray wolf optimization algorithm is used for optimizing the weight and bias parameters of the preset classifier so as to improve the detection performance of the target monitoring early warning model, in one implementation mode, the target sample data set and the target characteristic set can be input into the preset classifier, the target importance weight is determined based on the anchor point mechanism and the dynamic adjustment strategy in the preset classifier, the preset classifier is classified and trained according to the target importance weight, and the target monitoring early warning model is determined, wherein the target importance weight is the optimal weight configuration of the extreme learning machine classifier based on the improved anchor point gray wolf optimization algorithm, and the training classifier comprises the following steps: initialization, anchor point setting, fitness evaluation, updating strategy, adaptive feature importance feedback, iterative optimization and final model determination.
When the initializing step is carried out, the weight and bias parameters of the extreme learning machine classifier are required to be randomly initialized, meanwhile, the improved anchor point gray wolf population in the gray wolf optimization algorithm is initialized, and each gray wolf represents a set of parameter candidate solutions of the extreme learning machine classifier. Specifically, the weight and bias parameters of the extreme learning machine classifier are defined asAnd/>Each wolf in the improved anchor wolf optimization algorithm represents a set of possible/>And/>Values, i.e./>Wherein/>Representing the/>, in a populationOnly wolves. In one embodiment, the sirius population size is set to/>Maximum number of iterations/>。
When the anchor point setting step is carried out, an anchor point, namely a historical optimal solution, needs to be set in the improved anchor point gray wolf optimization algorithm, so that the search process is guided, and the stability and the convergence of the algorithm are enhanced. Specifically, an anchor pointIs historically the best/>The value can be randomly selected or set as the best/>, in the first generation。
In the step of fitness evaluation, for each gray wolf (i.e., each set of parameter candidate solutions) in the population, the classification accuracy of each gray wolf on the training data is calculated by a trained extreme learning machine classifier and used as a fitness function. Specifically, fitness functionFor evaluating each of the wolves/>The performance of the corresponding extreme learning machine parameters on the training data set, and in one embodiment, fitness is expressed in terms of classification accuracy:
Wherein, Is a training data set,/>A function is calculated for classification accuracy.
In the step of updating the strategy, the anchor points and the positions of the wolves in the population need to be updated according to the fitness of the wolves. The updating mechanism comprises following anchor points to explore and utilize, and dynamically adjusting the searching behavior of the wolf based on the current searching situation. Specifically, consider anchor pointsThe position update of the wolf follows the following rules:
Wherein, Is the best/>, in the current generation,/>And/>Is a random number used to control the impact of the anchor point and the current optimal solution on the gray wolf location update.
Further, random numbersAnd/>The specific calculation of (2) is as follows:
/>
Wherein, Generating a random number in the range of [0,1 ]/>Representing the current iteration number and the total iteration number/>And (3) the ratio of the search strategy is used for dynamically adjusting the search strategy to change the search behavior from global search to local search along with the iterative process.
When the step of self-adaptive feature importance feedback is carried out, the self-adaptive feature importance feedback module is required to dynamically evaluate and adjust the importance weights of all the features in the training process of the extreme learning machine, and the self-adaptive feature importance feedback can adaptively adjust the feature weights according to the performance and classification result of the current model, so as to highlight the features more important to classification tasks, unlike the traditional feature selection or weight distribution method. Specifically, it is provided withRepresents the/>Features, the initial importance weight of which is/>. In each iteration process, based on the classification performance of the current extreme learning machine model, the adaptive feature importance feedback module updates the importance weight of each feature:
Wherein, Is/>The update amount of the feature importance weights can be calculated according to the contribution degree of the features to the classification error samples:
Wherein, Is the misclassified sample set in the current iteration,/>Is/>First/>, of the samplesCharacteristic value/>Is/>Average of individual features over all training samples,/>Is a learning rate parameter, controls the magnitude of the weight update, in one embodiment,/>Set to 0.01.
In a particular embodiment, an initial feature importance weight is set1, Treating all the features equally, and as the iteration is carried out, dynamically adjusting the importance weights of the features by the self-adaptive feature importance feedback module according to the actual contribution of the features so as to better reflect the actual influence of the features on classification tasks.
When the steps of iterative optimization and final model determination are carried out, the steps of anchor point setting, fitness evaluation, updating strategy and adaptive feature importance feedback are required to be repeated until the maximum iteration number is reached, and the extreme learning machine classifier parameter corresponding to the gray wolf with the highest fitness is selected as the parameter of the final model, namely, the extreme learning machine classifier parameter with the highest fitness is selectedAs final parameters of the extreme learning machine classifier, it can be expressed as:
Further, after model training is completed, a target monitoring and early warning model can be used to process a new environmental protection index sample of the power plant and predict the possible environmental protection risk level, and in practical application, a new sample data is expressed as For new samples/>Extracting key features by applying a previously trained feature extraction model, wherein the extracted features are expressed as/>Further, extracted features/>Inputting the feature into a trained feature dimension reduction model to obtain feature representation/>Further, the feature after dimension reduction/>Inputting the new sample into a trained classifier to obtain the predicted category/>In one embodiment, the predicted category is 1 of normal emissions, out-of-standard emissions, equipment failure, and abnormal weather effects. /(I)
In summary, the invention can remarkably increase the size and diversity of the training data set by combining the generation of robustness constraint against the network, and is helpful to promote the generalization capability and accuracy of the model under the condition of limited original data, thereby effectively processing abnormal values and noise in the environmental monitoring data of the power plant, generating more real and diversified data samples and improving the effect of data expansion; the parameters of the neural network are optimized by adopting the regulation and control action of the simulated auxin in the growth and development of the plant, so that the learning efficiency and accuracy of the feature extraction model are improved, and particularly, when complex and high-dimensional data are processed, the good stability and performance can be kept, thereby effectively avoiding sinking into a local optimal solution and improving the global searching capability; by projecting the data into a stereo spiral space with exquisite design, the self-encoder not only learns the low-dimensional representation of the data in the encoding process, but also can enhance the capturing capability of the model on the internal structure of the data, and can more effectively retain the important information of the original data and remove redundancy and noise when the feature dimension reduction is carried out, thereby improving the performance of the subsequent classification model; by introducing an anchor point mechanism and a dynamic adjustment strategy, the improved anchor point gray wolf optimization algorithm enhances the searching capability and convergence speed of the algorithm, optimizes the parameters of the extreme learning machine classifier, improves the accuracy and generalization capability of the classifier, is particularly suitable for processing classification tasks of environmental protection monitoring data of a power plant, and can effectively improve the prediction capability of the model on environmental protection risk levels.
For the method for monitoring and early warning the environmental protection indexes of the power plant based on the artificial intelligence provided in the foregoing embodiment, the embodiment of the present invention provides a device for monitoring and early warning the environmental protection indexes of the power plant based on the artificial intelligence, referring to a schematic structural diagram of the device for monitoring and early warning the environmental protection indexes of the power plant based on the artificial intelligence shown in fig. 3, the device comprises the following parts:
The sample data acquisition module 302 determines the power plant environment monitoring data sent by the power plant data sensor as an original sample data set;
The data expansion module 304 is used for carrying out data expansion processing on the original sample data set after enhancing the robustness of noise and abnormal values in the original sample data set by enhancing the generated countermeasure network algorithm through the robustness enhanced generation countermeasure network algorithm, so as to determine a target sample data set;
The model training module 306 performs model training processing on a preset power plant environmental protection index monitoring and early warning model based on the target sample data set to determine a target monitoring and early warning model, wherein the target monitoring and early warning model is used for performing index monitoring processing on the power plant environmental protection index to determine a monitoring and early warning result.
The power plant environmental protection index monitoring and early warning device based on the artificial intelligence provided by the embodiment of the application can obviously improve the accuracy of power plant environmental protection index monitoring and early warning.
In one embodiment, when performing the step of performing model training processing on the preset power plant environmental protection index monitoring and early warning model based on the target sample data set and determining the target monitoring and early warning model, the model training module 306 is further configured to: performing feature extraction processing on a target sample data set based on a neural network parameter optimization algorithm of auxin regulation optimization, and determining a feature set; inputting the target sample data set and the feature set into a preset classifier, and performing classification training on the preset classifier based on an improved anchor point and wolf optimization algorithm to determine a target monitoring and early warning model, wherein the improved anchor point and wolf optimization algorithm is used for optimizing the weight and bias parameters of the preset classifier so as to improve the detection performance of the target monitoring and early warning model.
In one embodiment, when performing the neural network parameter optimization algorithm based on auxin regulation optimization, performing feature extraction processing on the target sample data set, and determining the feature set, the model training module 306 is further configured to: adjusting parameters of the neural network by using a neural network parameter optimization algorithm based on auxin regulation and optimization and simulating a distribution and migration mechanism of auxin so as to perform parameter optimization adjustment processing on weights and biases of the neural network and determine a target neural network; and performing feature extraction processing on the target sample data set by using the target neural network model to determine a feature set.
In one embodiment, when performing the step of using the neural network parameter optimization algorithm based on auxin regulation optimization to adjust the neural network parameters by simulating the distribution and migration mechanism of auxin to perform parameter optimization adjustment on the weights and biases of the neural network, the model training module 306 is further configured to: simulating the distribution condition of initialized auxin in a preset parameter space, wherein each parameter vector in the preset parameter space is used for simulating the cell position in a plant body, the corresponding fitness of each parameter vector is used for simulating the auxin concentration of the cell position, the parameter vector is used for representing the weight and bias of a neural network, and the fitness is positively related to the auxin concentration; and (3) performing parameter optimization adjustment processing on the neural network parameters by simulating the migration process of auxin between cells from a high-concentration auxin region to a low-concentration auxin region, and determining the target neural network.
In one embodiment, when the target sample data set and the feature set are input into the preset classifier, and the preset classifier is classified and trained based on the improved anchor point gray wolf optimization algorithm, and the target monitoring and early warning model is determined, the model training module 306 is further configured to: inputting the feature set into a preset feature dimension reduction model, performing data noise reduction processing on the feature set by using a three-dimensional spiral projection self-encoder in the preset feature dimension reduction model, and determining a target feature set after noise reduction.
In one embodiment, when the step of inputting the feature set into the preset feature dimension reduction model and performing data noise reduction processing on the feature set by using the stereo spiral projection self-encoder in the preset feature dimension reduction model and determining the target feature set after noise reduction, the model training module 306 is further configured to: projecting the feature set to a stereoscopic spiral space through a stereoscopic spiral projection self-encoder, and carrying out parameter adjustment processing on the shape and the size of the spiral according to feature dynamics corresponding to each data in the stereoscopic spiral space to determine a target stereoscopic spiral space, wherein the feature dynamics comprise: radius features, helix angle features, angular rotation rate features, and height features along the helical axis; and in the target stereo spiral space, performing feature dimension reduction processing on the projected data by using a preset low-dimensional coding learning model to remove redundant information and noise, and reserving a target feature set containing key information.
In one embodiment, when the step of inputting the target sample data set and the feature set into the preset classifier, and performing classification training on the preset classifier based on the improved anchor point gray wolf optimization algorithm, and determining the target monitoring and early warning model, the model training module 306 is further configured to: inputting a target sample data set and a target feature set into a preset classifier, and determining a target importance weight based on an anchor point mechanism and a dynamic adjustment strategy through an anchor point gray wolf optimization algorithm based on improvement in the preset classifier, wherein the target importance weight is the optimal weight configuration of an extreme learning machine classifier based on the anchor point gray wolf optimization algorithm; and carrying out classification training on a preset classifier according to the importance weight of the target, and determining a target monitoring and early warning model.
The device provided by the embodiment of the present invention has the same implementation principle and technical effects as those of the foregoing method embodiment, and for the sake of brevity, reference may be made to the corresponding content in the foregoing method embodiment where the device embodiment is not mentioned.
The embodiment of the invention provides electronic equipment, which comprises a processor and a storage device; the storage means has stored thereon a computer program which, when executed by the processor, performs the method of any of the embodiments described above.
Fig. 4 is a schematic structural diagram of an electronic device according to an embodiment of the present invention, where the electronic device 100 includes: a processor 40, a memory 41, a bus 42 and a communication interface 43, the processor 40, the communication interface 43 and the memory 41 being connected by the bus 42; the processor 40 is arranged to execute executable modules, such as computer programs, stored in the memory 41.
The memory 41 may include a high-speed random access memory (RAM, random Access Memory), and may further include a non-volatile memory (non-volatilememory), such as at least one magnetic disk memory. The communication connection between the system network element and the at least one other network element is achieved via at least one communication interface 43 (which may be wired or wireless), which may use the internet, a wide area network, a local network, a metropolitan area network, etc.
Bus 42 may be an ISA bus, a PCI bus, an EISA bus, or the like. The buses may be classified as address buses, data buses, control buses, etc. For ease of illustration, only one bi-directional arrow is shown in FIG. 4, but not only one bus or type of bus.
The memory 41 is configured to store a program, and the processor 40 executes the program after receiving an execution instruction, and the method executed by the apparatus for flow defining disclosed in any of the foregoing embodiments of the present invention may be applied to the processor 40 or implemented by the processor 40.
The processor 40 may be an integrated circuit chip having signal processing capabilities. In implementation, the steps of the above method may be performed by integrated logic circuitry in hardware or instructions in software in processor 40. The processor 40 may be a general-purpose processor, including a Central Processing Unit (CPU), a network processor (NetworkProcessor NP), etc.; but may also be a digital signal processor (DIGITAL SIGNAL Processing, DSP), application SPECIFIC INTEGRATED Circuit (ASIC), off-the-shelf Programmable gate array (Field-Programmable GATE ARRAY, FPGA) or other Programmable logic device, discrete gate or transistor logic device, discrete hardware components. The disclosed methods, steps, and logic blocks in the embodiments of the present invention may be implemented or performed. A general purpose processor may be a microprocessor or the processor may be any conventional processor or the like. The steps of the method disclosed in connection with the embodiments of the present invention may be embodied directly in the execution of a hardware decoding processor, or in the execution of a combination of hardware and software modules in a decoding processor. The software modules may be located in a random access memory, flash memory, read only memory, programmable read only memory, or electrically erasable programmable memory, registers, etc. as well known in the art. The storage medium is located in a memory 41 and the processor 40 reads the information in the memory 41 and in combination with its hardware performs the steps of the method described above.
The computer program product of the readable storage medium provided by the embodiment of the present invention includes a computer readable storage medium storing a program code, where the program code includes instructions for executing the method described in the foregoing method embodiment, and the specific implementation may refer to the foregoing method embodiment and will not be described herein.
The functions, if implemented in the form of software functional units and sold or used as a stand-alone product, may be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention may be embodied essentially or in a part contributing to the prior art or in a part of the technical solution, in the form of a software product stored in a storage medium, comprising several instructions for causing a computer device (which may be a personal computer, an electronic device, or a network device, etc.) to perform all or part of the steps of the method according to the embodiments of the present invention. And the aforementioned storage medium includes: a usb disk, a removable hard disk, a Read-Only Memory (ROM), a random access Memory (RAM, random Access Memory), a magnetic disk, or an optical disk, or other various media capable of storing program codes.
Finally, it should be noted that: the above examples are only specific embodiments of the present invention, and are not intended to limit the scope of the present invention, but it should be understood by those skilled in the art that the present invention is not limited thereto, and that the present invention is described in detail with reference to the foregoing examples: any person skilled in the art may modify or easily conceive of the technical solution described in the foregoing embodiments, or perform equivalent substitution of some of the technical features, while remaining within the technical scope of the present disclosure; such modifications, changes or substitutions do not depart from the spirit and scope of the technical solutions of the embodiments of the present invention, and are intended to be included in the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (10)
1. An artificial intelligence-based power plant environmental protection index monitoring and early warning method is characterized by comprising the following steps:
Determining the power plant environment monitoring data sent by a power plant data sensor as an original sample data set;
Generating an countermeasure network algorithm through the robustness enhancement, after the robustness of noise and abnormal values in the original sample data set is enhanced by the generated countermeasure network algorithm, performing data expansion processing on the original sample data set, and determining a target sample data set;
And carrying out model training treatment on a preset power plant environmental protection index monitoring and early warning model based on the target sample data set to determine a target monitoring and early warning model, wherein the target monitoring and early warning model is used for carrying out index monitoring treatment on the power plant environmental protection index to determine a monitoring and early warning result.
2. The method for monitoring and early warning of environmental indicators of a power plant based on artificial intelligence according to claim 1, wherein the step of performing model training processing on a preset monitoring and early warning model of environmental indicators of the power plant based on the target sample data set to determine the target monitoring and early warning model comprises the following steps:
Performing feature extraction processing on the target sample data set based on a neural network parameter optimization algorithm of auxin regulation optimization, and determining a feature set;
Inputting the target sample data set and the feature set into a preset classifier, and performing classification training on the preset classifier based on an improved anchor point gray wolf optimization algorithm to determine the target monitoring early warning model, wherein the improved anchor point gray wolf optimization algorithm is used for optimizing the weight and bias parameters of the preset classifier so as to improve the detection performance of the target monitoring early warning model.
3. The method for monitoring and early warning environmental protection indexes of a power plant based on artificial intelligence according to claim 2, wherein the step of performing feature extraction processing on the target sample data set and determining a feature set by using the neural network parameter optimization algorithm based on auxin regulation optimization comprises the following steps:
Adjusting the parameters of the neural network by using the neural network parameter optimization algorithm based on auxin regulation and optimization and simulating the distribution and migration mechanism of auxin so as to perform parameter optimization adjustment processing on the weights and the biases of the neural network and determine a target neural network;
and carrying out feature extraction processing on the target sample data set by using the target neural network to determine a feature set.
4. The method for monitoring and early warning environmental indicators of a power plant based on artificial intelligence according to claim 3, wherein the step of determining the target neural network by using the neural network parameter optimization algorithm based on auxin regulation and optimization to adjust the neural network parameters by simulating the distribution and migration mechanism of auxin to perform parameter optimization adjustment processing on the weights and biases of the neural network comprises the following steps:
Simulating the distribution condition of initialized auxin in a preset parameter space, wherein each parameter vector in the preset parameter space is used for simulating the cell position in a plant body, the fitness corresponding to each parameter vector is used for simulating the auxin concentration of the cell position, the parameter vector is used for representing the weight and bias of a neural network, and the fitness is positively related to the auxin concentration;
and carrying out parameter optimization adjustment treatment on the neural network parameters by simulating the migration process of auxin between cells from a high-concentration auxin region to a low-concentration auxin region, and determining the target neural network.
5. The method for monitoring and early warning of environmental indicators of a power plant based on artificial intelligence according to claim 2, wherein the step of inputting the target sample data set and the feature set into a preset classifier, and performing classification training on the preset classifier based on an improved anchor point gray wolf optimization algorithm, and determining the target monitoring and early warning model comprises the following steps:
Inputting the feature set into a preset feature dimension reduction model, performing data noise reduction processing on the feature set by using a three-dimensional spiral projection self-encoder in the preset feature dimension reduction model, and determining a target feature set after noise reduction.
6. The method for monitoring and early warning environmental indicators of a power plant based on artificial intelligence according to claim 5, wherein the step of inputting the feature set into a preset feature dimension reduction model, performing data noise reduction processing on the feature set by using a stereo spiral projection self-encoder in the preset feature dimension reduction model, and determining a target feature set after noise reduction comprises the following steps:
Projecting the feature set to a stereoscopic spiral space through a stereoscopic spiral projection self-encoder, and carrying out parameter adjustment processing on the shape and the size of the spiral according to feature dynamics corresponding to each data in the stereoscopic spiral space to determine a target stereoscopic spiral space, wherein the feature dynamics comprise: radius features, helix angle features, angular rotation rate features, and height features along the helical axis;
And in the target stereo spiral space, performing feature dimension reduction processing on the projected data by using a preset low-dimensional coding learning model to remove redundant information and noise, and reserving the target feature set containing key information.
7. The method for monitoring and early warning of environmental indicators of a power plant based on artificial intelligence according to claim 2, wherein the steps of inputting the target sample data set and the feature set into a preset classifier, classifying and training the preset classifier based on an improved anchor point gray wolf optimization algorithm, and determining the target monitoring and early warning model comprise the following steps:
inputting the target sample data set and the target feature set into a preset classifier, and determining a target importance weight based on an anchor point mechanism and a dynamic adjustment strategy through an improved anchor point gray wolf optimization algorithm in the preset classifier, wherein the target importance weight is the optimal weight configuration of an extreme learning machine classifier based on the improved anchor point gray wolf optimization algorithm;
and carrying out classification training on the preset classifier according to the target importance weight, and determining the target monitoring and early warning model.
8. An artificial intelligence-based power plant environmental protection index monitoring and early warning device is characterized in that the device comprises:
The sample data acquisition module is used for determining the power plant environment monitoring data sent by the power plant data sensor as an original sample data set;
The data expansion module is used for carrying out data expansion processing on the original sample data set after enhancing the robustness of noise and abnormal values in the original sample data set through a robust enhanced generation countermeasure network algorithm and enhancing the robustness of the generation countermeasure network, so as to determine a target sample data set;
The model training module is used for carrying out model training processing on a preset power plant environmental protection index monitoring and early warning model based on the target sample data set to determine a target monitoring and early warning model, wherein the target monitoring and early warning model is used for carrying out index monitoring processing on the power plant environmental protection index to determine a monitoring and early warning result.
9. An electronic device comprising a processor and a memory, the memory storing computer-executable instructions executable by the processor, the processor executing the computer-executable instructions to implement the method of any one of claims 1 to 7.
10. A computer readable storage medium storing computer executable instructions which, when invoked and executed by a processor, cause the processor to implement the method of any one of claims 1 to 7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410571507.6A CN118133002B (en) | 2024-05-10 | Power plant environment-friendly index monitoring and early warning method and device based on artificial intelligence |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202410571507.6A CN118133002B (en) | 2024-05-10 | Power plant environment-friendly index monitoring and early warning method and device based on artificial intelligence |
Publications (2)
Publication Number | Publication Date |
---|---|
CN118133002A true CN118133002A (en) | 2024-06-04 |
CN118133002B CN118133002B (en) | 2024-07-30 |
Family
ID=
Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113378988A (en) * | 2021-07-06 | 2021-09-10 | 浙江工业大学 | Deep learning system robustness enhancement method and device based on particle swarm optimization |
CN115018191A (en) * | 2022-06-29 | 2022-09-06 | 同济大学 | Carbon emission prediction method based on small sample data |
CN115201608A (en) * | 2022-07-26 | 2022-10-18 | 广东粤电靖海发电有限公司 | Power plant equipment operation parameter monitoring method based on neural network |
CN117290732A (en) * | 2023-11-24 | 2023-12-26 | 山东理工昊明新能源有限公司 | Construction method of fault classification model, wind power equipment fault classification method and device |
CN117312865A (en) * | 2023-11-30 | 2023-12-29 | 山东理工职业学院 | Nonlinear dynamic optimization-based data classification model construction method and device |
CN117892251A (en) * | 2024-03-18 | 2024-04-16 | 山东神力索具有限公司 | Rigging forging process parameter monitoring and early warning method and device based on artificial intelligence |
CN117892182A (en) * | 2024-03-14 | 2024-04-16 | 山东神力索具有限公司 | Rope durability testing method and device based on artificial intelligence |
CN117992899A (en) * | 2024-04-07 | 2024-05-07 | 国网山东省电力公司蒙阴县供电公司 | Small-sample-based fault diagnosis method and device for new electric energy equipment |
Patent Citations (8)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN113378988A (en) * | 2021-07-06 | 2021-09-10 | 浙江工业大学 | Deep learning system robustness enhancement method and device based on particle swarm optimization |
CN115018191A (en) * | 2022-06-29 | 2022-09-06 | 同济大学 | Carbon emission prediction method based on small sample data |
CN115201608A (en) * | 2022-07-26 | 2022-10-18 | 广东粤电靖海发电有限公司 | Power plant equipment operation parameter monitoring method based on neural network |
CN117290732A (en) * | 2023-11-24 | 2023-12-26 | 山东理工昊明新能源有限公司 | Construction method of fault classification model, wind power equipment fault classification method and device |
CN117312865A (en) * | 2023-11-30 | 2023-12-29 | 山东理工职业学院 | Nonlinear dynamic optimization-based data classification model construction method and device |
CN117892182A (en) * | 2024-03-14 | 2024-04-16 | 山东神力索具有限公司 | Rope durability testing method and device based on artificial intelligence |
CN117892251A (en) * | 2024-03-18 | 2024-04-16 | 山东神力索具有限公司 | Rigging forging process parameter monitoring and early warning method and device based on artificial intelligence |
CN117992899A (en) * | 2024-04-07 | 2024-05-07 | 国网山东省电力公司蒙阴县供电公司 | Small-sample-based fault diagnosis method and device for new electric energy equipment |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN108197736B (en) | Air quality prediction method based on variational self-encoder and extreme learning machine | |
CN117851921B (en) | Equipment life prediction method and device based on transfer learning | |
CN118032327B (en) | Equipment intelligent lubrication monitoring method and device based on artificial intelligence | |
CN117892251A (en) | Rigging forging process parameter monitoring and early warning method and device based on artificial intelligence | |
CN116402203A (en) | Method, system and medium for predicting short-time photovoltaic power generation capacity considering weather conditions | |
CN115952456A (en) | Method, system, program product and storage medium for determining fault diagnosis model | |
CN113762591B (en) | Short-term electric quantity prediction method and system based on GRU and multi-core SVM countermeasure learning | |
CN117290732B (en) | Construction method of fault classification model, wind power equipment fault classification method and device | |
CN118133002B (en) | Power plant environment-friendly index monitoring and early warning method and device based on artificial intelligence | |
CN118133002A (en) | Power plant environment-friendly index monitoring and early warning method and device based on artificial intelligence | |
CN116933090A (en) | PM2.5 prediction method and device based on deep neural network | |
CN115423091A (en) | Conditional antagonistic neural network training method, scene generation method and system | |
CN118134058B (en) | Power plant carbon emission accounting method, device and server based on big data | |
CN118171180B (en) | Equipment state prediction method and device based on artificial intelligence | |
CN118133689B (en) | Teaching scene-oriented simulation platform data processing method and device and electronic equipment | |
CN117892182B (en) | Rope durability testing method and device based on artificial intelligence | |
CN118070682B (en) | Spiral bolt hanging ring damage assessment method and device based on artificial intelligence | |
CN115100322B (en) | Line element self-adaptive simplification method and device for supervised learning supported multi-reduction algorithm cooperation | |
CN118070134B (en) | Method and system for predicting electricity consumption of users in park | |
CN117312865B (en) | Nonlinear dynamic optimization-based data classification model construction method and device | |
CN116992944B (en) | Image processing method and device based on leavable importance judging standard pruning | |
CN112161621B (en) | Model-free auxiliary navigation adaptive area selection method | |
CN118332388A (en) | Fault detection method and device of photovoltaic cleaning equipment and electronic equipment | |
CN118012182A (en) | Control method, system, equipment and storage medium for operating environment of high-voltage switch cabinet | |
CN117290673A (en) | Ship energy consumption high-precision prediction system based on multi-model fusion |
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 |