Thermal power generating unit operation initial pressure optimization method and device

Abstract

The invention provides a thermal power generating unit operation initial pressure optimization method and device, and relates to the technical field of optimization control. The method comprises the following steps: acquiring working parameters of a thermal power generating unit in a use stage; and processing the working parameters in the use stage based on an optimized random forest model to obtain the main steam pressure flow when the heat consumption rate reaches the minimum value, and taking the main steam pressure flow at the moment as the initial pressure of operation of the thermal power generating unit. The device executes the method. According to the thermal power generating unit operation initial pressure optimization method and device provided by the embodiment of the invention, the response speed and the control precision of thermal power generating unit operation initial pressure control can be improved.

Description

Technical Field

[0001] The present invention relates to the technical field of optimization control, and in particular to a method and device for optimizing the initial pressure of a thermal power unit. Background Art

[0002] With the significant increase in installed power generation capacity, most thermal power units are operated in a sliding pressure control mode. In order to maintain the thermal economy of thermal power units, it is necessary to synchronously control the initial operating pressure of thermal power units. In the prior art, the control of the initial operating pressure of thermal power units is usually carried out by theoretical optimization or experimental optimization. Such control strategies have the problems of slow response speed and poor control accuracy. Summary of the invention

[0003] In view of the problems in the prior art, an embodiment of the present invention provides a method and device for optimizing the initial pressure of a thermal power unit, which can at least partially solve the problems in the prior art.

[0004] On the one hand, the present invention provides a method for optimizing the initial pressure of a thermal power unit, comprising:

[0005] Obtain the working parameters of the thermal power unit during the use phase;

[0006] The working parameters of the use phase are processed based on the optimized random forest model to obtain the main steam pressure flow when the heat rate reaches the minimum value, and the main steam pressure flow at this time is used as the initial operating pressure of the thermal power unit;

[0007] The optimized random forest model includes feedback trees corresponding to each decision tree one by one, and is obtained by training after linear normalization of working parameters in the training phase according to a dynamic vector field; the topological structure of the nodes in each feedback tree is mirror-symmetrical to the topological structure of the nodes in the corresponding decision tree, and each feedback tree includes a comparison logic node connected to each node in the feedback tree, and the comparison logic node has a topological connection relationship with the corresponding comparison logic nodes in other feedback trees.

[0008] The linear normalization processing of the working parameters in the training phase according to the dynamic vector field includes:

[0009] Arranging the working parameters of the training phase in a time sequence to obtain a working parameter matrix;

[0010] Mapping the matrix elements in the working parameter matrix to the dynamic vector field to obtain a working parameter vector set;

[0011] According to the linear correlation dimension of each working parameter vector in the working parameter vector set, a first-order working parameter dynamic vector value is obtained, and the first-order working parameter dynamic vector value is used as the weight corresponding to each working parameter vector to perform linear transformation on the working parameter matrix to achieve linear normalization processing.

[0012] The step of obtaining the first-order working parameter dynamic vector value according to the linear correlation dimension of each working parameter vector in the working parameter vector set includes:

[0013] According to the value of the linear correlation dimension of each working parameter vector in the working parameter vector set, a corresponding single or multiple differential calculation is performed on each working parameter vector to obtain the first-order working parameter dynamic vector value.

[0014] Wherein, training the optimized random forest model includes:

[0015] According to the training phase operating parameters after linear normalization processing, a correlation data set between the training phase operating parameters and the heat rate is obtained, and the correlation data set is used as the training data set;

[0016] Extracting features from the training data set to obtain a feature value set, and using the feature value set as the best segmentation feature to recursively split the decision tree to obtain multiple decision trees;

[0017] Inputting the working parameters of the training phase into each decision tree, outputting the main steam pressure flow value corresponding to the minimum heat rate through each decision tree, and inputting the main steam pressure flow value into the feedback tree corresponding to the decision tree;

[0018] The feedback result output by each feedback tree is obtained, and the deviation value between each feedback result and the working parameter of the training stage is corrected until a preset iteration termination condition is met.

[0019] The step of obtaining the feedback result output by each feedback tree includes:

[0020] Traversing the nodes in each feedback tree respectively to perform reverse judgment on the main steam pressure flow value, and when each main steam pressure flow judgment value corresponding to a node is obtained, the main steam pressure flow judgment value is transmitted to the comparison logic node in the corresponding feedback tree, and the main steam pressure flow judgment value in the current comparison logic node is compared with the main steam pressure flow judgment values ​​in other comparison logic nodes that have a topological connection relationship with the current comparison logic node;

[0021] If it is determined that the preset comparison logic relationship is met, continue to traverse the next node in the current feedback tree in the reverse judgment order, and repeat the main steam pressure flow judgment value corresponding to each node and subsequent steps until all nodes in the current feedback tree are traversed, and the working parameters corresponding to the main steam pressure flow judgment value at this time are used as the feedback results.

[0022] Wherein, the method for optimizing the initial pressure of thermal power unit operation further comprises:

[0023] If it is determined that the preset comparison logic relationship is not met, the main steam pressure flow determination value is deleted.

[0024] Wherein, the preset termination condition includes that the number of repeated feedback results reaches a preset upper limit number; accordingly, the method for optimizing the initial pressure of the thermal power unit operation further includes:

[0025] When the number of repeated feedback results reaches a preset upper limit, the main steam pressure flow in the iteration round with the smallest deviation from the working parameters in the training stage is selected as the initial operating pressure of the thermal power unit when the training is terminated.

[0026] On the one hand, the present invention provides a device for optimizing the initial pressure of a thermal power unit, comprising:

[0027] An acquisition unit, used for acquiring the working parameters of the thermal power unit during the use phase;

[0028] An optimization unit is used to process the working parameters of the use phase based on an optimization random forest model to obtain the main steam pressure flow when the heat rate reaches a minimum value, and use the main steam pressure flow at this time as the initial operating pressure of the thermal power unit;

[0029] The optimized random forest model includes feedback trees corresponding to each decision tree one by one, and is obtained by training after linear normalization of working parameters in the training phase according to a dynamic vector field; the topological structure of the nodes in each feedback tree is mirror-symmetrical to the topological structure of the nodes in the corresponding decision tree, and each feedback tree includes a comparison logic node connected to each node in the feedback tree, and the comparison logic node has a topological connection relationship with the corresponding comparison logic nodes in other feedback trees.

[0030] In another aspect, an embodiment of the present invention provides an electronic device, including: a processor, a memory, and a bus, wherein:

[0031] The processor and the memory communicate with each other via the bus;

[0032] The memory stores program instructions that can be executed by the processor, and the processor calls the program instructions to execute the following method:

[0033] Obtain the working parameters of the thermal power unit during the use phase;

[0034] The working parameters of the use phase are processed based on the optimized random forest model to obtain the main steam pressure flow when the heat rate reaches the minimum value, and the main steam pressure flow at this time is used as the initial operating pressure of the thermal power unit;

[0035] The optimized random forest model includes feedback trees corresponding to each decision tree one by one, and is obtained by training after linear normalization of working parameters in the training phase according to a dynamic vector field; the topological structure of the nodes in each feedback tree is mirror-symmetrical to the topological structure of the nodes in the corresponding decision tree, and each feedback tree includes a comparison logic node connected to each node in the feedback tree, and the comparison logic node has a topological connection relationship with the corresponding comparison logic nodes in other feedback trees.

[0036] An embodiment of the present invention provides a non-transitory computer-readable storage medium, including:

[0037] The non-transitory computer-readable storage medium stores computer instructions, which cause the computer to execute the following method:

[0038] Obtain the working parameters of the thermal power unit during the use phase;

[0039] The working parameters of the use phase are processed based on the optimized random forest model to obtain the main steam pressure flow when the heat rate reaches the minimum value, and the main steam pressure flow at this time is used as the initial operating pressure of the thermal power unit;

[0040] The optimized random forest model includes feedback trees corresponding to each decision tree one by one, and is obtained by training after linear normalization of working parameters in the training phase according to a dynamic vector field; the topological structure of the nodes in each feedback tree is mirror-symmetrical to the topological structure of the nodes in the corresponding decision tree, and each feedback tree includes a comparison logic node connected to each node in the feedback tree, and the comparison logic node has a topological connection relationship with the corresponding comparison logic nodes in other feedback trees.

[0041] The method and device for optimizing the initial operating pressure of a thermal power unit provided in an embodiment of the present invention obtain the operating parameters of the thermal power unit in the use phase; process the operating parameters in the use phase based on an optimized random forest model to obtain the main steam pressure flow when the heat rate reaches the minimum value, and use the main steam pressure flow at this time as the initial operating pressure of the thermal power unit; wherein the optimized random forest model includes a feedback tree corresponding to each decision tree one by one, and is trained after linear normalization processing is performed on the working parameters in the training phase according to a dynamic vector field; the topological structure of the nodes in each feedback tree is mirror-symmetric to the topological structure of the nodes in the corresponding decision tree, and each feedback tree respectively includes a comparison logic node connected to each node in the feedback tree, and the comparison logic node has a topological connection relationship with the comparison logic nodes corresponding to each other in other feedback trees, which can improve the response speed and control accuracy of the initial operating pressure control of the thermal power unit. BRIEF DESCRIPTION OF THE DRAWINGS

[0042] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the following briefly introduces the drawings required for use in the embodiments or the prior art descriptions. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0043] Figure 1 It is a flow chart of a method for optimizing the initial pressure of a thermal power unit provided in one embodiment of the present invention.

[0044] Figure 2 It is a structural schematic diagram of an initial pressure optimization device for a thermal power unit provided in one embodiment of the present invention.

[0045] Figure 3 A schematic diagram of the physical structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0046] In order to make the purpose, technical scheme and advantages of the embodiments of the present invention more clear, the embodiments of the present invention are further described in detail below in conjunction with the accompanying drawings. Here, the illustrative embodiments of the present invention and their descriptions are used to explain the present invention, but are not intended to limit the present invention. It should be noted that, in the absence of conflict, the embodiments in this application and the features in the embodiments can be combined with each other arbitrarily.

[0047] Figure 1 FIG. 1 is a flow chart of a method for optimizing the initial pressure of a thermal power unit according to an embodiment of the present invention. Figure 1 As shown, the method for optimizing the initial pressure of a thermal power unit provided by the embodiment of the present invention includes:

[0048] Step S1: Obtaining the operating parameters of the thermal power unit during the use phase.

[0049] Step S2: processing the working parameters of the use phase based on the optimized random forest model to obtain the main steam pressure flow when the heat rate reaches the minimum value, and using the main steam pressure flow at this time as the initial operating pressure of the thermal power unit;

[0050] The optimized random forest model includes feedback trees corresponding to each decision tree one by one, and is obtained by training after linear normalization of working parameters in the training phase according to a dynamic vector field; the topological structure of the nodes in each feedback tree is mirror-symmetrical to the topological structure of the nodes in the corresponding decision tree, and each feedback tree includes a comparison logic node connected to each node in the feedback tree, and the comparison logic node has a topological connection relationship with the corresponding comparison logic nodes in other feedback trees.

[0051] In the above step S1, the device obtains the working parameters of the thermal power unit in the use phase. The device can be a computer device that executes the method, etc. The working parameters in the use phase are the same as the working parameters in the subsequent training phase, except that the use phase is different, and are collectively referred to as working parameters, specifically including the pressure before the stage group, the pressure after the adjustment stage, the first stage exhaust temperature, and the high-pressure cylinder exhaust pressure.

[0052] In the above step S2, the device processes the working parameters of the use phase based on the optimized random forest model to obtain the main steam pressure flow when the heat rate reaches the minimum value, and uses the main steam pressure flow at this time as the initial operating pressure of the thermal power unit; the working parameters of the use phase can be input into the optimized random forest model, and the output result of the optimized random forest model can be used as the main steam pressure flow when the heat rate reaches the minimum value.

[0053] The optimized random forest model includes feedback trees corresponding to each decision tree one by one, and is obtained by training after linear normalization of working parameters in the training phase according to a dynamic vector field; the topological structure of the nodes in each feedback tree is mirror-symmetrical to the topological structure of the nodes in the corresponding decision tree, and each feedback tree includes a comparison logic node connected to each node in the feedback tree, and the comparison logic node has a topological connection relationship with the corresponding comparison logic nodes in other feedback trees.

[0054] The existing working parameters have the characteristics of multi-dimensional large data volume, which leads to the problem that the existing random forest model has untimely output response and is prone to overfitting when processing the above working parameters. The working parameters of the training phase can be improved first, and the working parameters of the training phase are linearly normalized according to the dynamic vector field, including:

[0055] Arranging the working parameters of the training phase in a time sequence to obtain a working parameter matrix;

[0056] Mapping the matrix elements in the working parameter matrix to the dynamic vector field to obtain a working parameter vector set;

[0057] According to the linear correlation dimension of each working parameter vector in the working parameter vector set, a first-order working parameter dynamic vector value is obtained, and the first-order working parameter dynamic vector value is used as the weight corresponding to each working parameter vector to perform linear transformation on the working parameter matrix to achieve linear normalization processing.

[0058] The obtaining of the first-order working parameter dynamic vector value according to the linear correlation dimension of each working parameter vector in the working parameter vector set includes:

[0059] According to the value of the linear correlation dimension of each working parameter vector in the working parameter vector set, a corresponding single or multiple differential calculation is performed on each working parameter vector to obtain the first-order working parameter dynamic vector value.

[0060] The instructions are as follows:

[0061] The working parameters of the thermal power units are organized into several time series vectors according to the time series, each time series vector includes all the working parameters of the thermal power units in the current time series, and the time series vectors are arranged in the time series order to obtain a working parameter matrix. The working parameter matrix can be an n×m matrix, wherein n represents n acquisition time points, m represents m types of working parameters of the thermal power units, and each matrix element represents the working parameter value of a certain type of working parameter of the thermal power unit at the corresponding acquisition time point.

[0062] Construct a dynamic vector field of the same dimension as the working parameter matrix, map the working parameter matrix to the dynamic vector field, obtain a working parameter vector set, perform single or multiple differentiation on each working parameter vector in its linearly related dimension, and obtain a first-order working parameter dynamic vector; specifically, if the linearly related dimension is one-dimensional, a single differentiation calculation can be used to obtain a first-order working parameter dynamic vector value; if the linearly related dimension is two-dimensional, a double differentiation calculation can be used to obtain a first-order working parameter dynamic vector value, and so on, which will not be repeated.

[0063] The first-order working parameter dynamic vector value at the corresponding moment of each working parameter is used as a weight to perform linear transformation on the working parameter matrix to complete normalization. After the working parameter matrix is ​​normalized, it is associated with the main steam pressure flow through linear regression. By preprocessing the working parameters of the thermal power unit, its nonlinearity is reduced, and it has obvious reversible symmetry with the main steam pressure flow, the output response speed of the model can be improved, and it is more suitable for the optimized random forest model obtained by the improvement of the present invention.

[0064] The model structure of the optimized random forest model is described as follows:

[0065] Each decision tree in the existing random forest model constructs a one-to-one corresponding feedback tree, and the topological structure of the nodes in each feedback tree is mirror-symmetric to the topological structure of the nodes in the corresponding decision tree (that is, the decision tree traverses each node in turn according to the positive judgment order from the root node to the leaf node from top to bottom, and the feedback tree traverses each node in turn according to the reverse judgment order from the leaf node to the root node from bottom to top), and the nodes of the feedback tree contain the reverse judgment logic of the nodes in the decision tree that are mirror-symmetric to the feedback tree. Each feedback tree contains a comparison logic node, which is connected to each node in the feedback tree and is used to obtain the main steam pressure flow judgment value so as to compare it with the main steam pressure flow judgment value in the comparison logic nodes on other feedback trees. There is a topological connection relationship between the comparison logic nodes of the above-mentioned other feedback trees and the comparison logic nodes of the current feedback tree, and the above-mentioned topological connection relationship can be the same as the topological connection relationship between the nodes of the corresponding decision trees.

[0066] Training the optimized random forest model includes:

[0067] According to the training phase operating parameters after linear normalization processing, a correlation data set between the training phase operating parameters and the heat rate is obtained, and the correlation data set is used as the training data set;

[0068] Extracting features from the training data set to obtain a feature value set, and using the feature value set as the best segmentation feature to recursively split the decision tree to obtain multiple decision trees;

[0069] Inputting the working parameters of the training phase into each decision tree, outputting the main steam pressure flow value corresponding to the minimum heat rate through each decision tree, and inputting the main steam pressure flow value into the feedback tree corresponding to the decision tree;

[0070] The feedback result output by each feedback tree is obtained, and the deviation value between each feedback result and the working parameter of the training stage is corrected until a preset iteration termination condition is met.

[0071] The obtaining of the feedback result output by each feedback tree includes:

[0072] Traversing the nodes in each feedback tree respectively to perform reverse judgment on the main steam pressure flow value, and when each main steam pressure flow judgment value corresponding to a node is obtained, the main steam pressure flow judgment value is transmitted to the comparison logic node in the corresponding feedback tree, and the main steam pressure flow judgment value in the current comparison logic node is compared with the main steam pressure flow judgment values ​​in other comparison logic nodes that have a topological connection relationship with the current comparison logic node;

[0073] If it is determined that the preset comparison logic relationship is met, continue to traverse the next node in the current feedback tree in the reverse judgment order, and repeat the main steam pressure flow judgment value corresponding to each node and subsequent steps until all nodes in the current feedback tree are traversed, and the working parameters corresponding to the main steam pressure flow judgment value at this time are used as the feedback results.

[0074] The method for optimizing the initial operating pressure of a thermal power unit further comprises:

[0075] If it is determined that the preset comparison logic relationship is not met, the main steam pressure flow determination value is deleted.

[0076] The preset termination condition includes that the number of repeated feedback results reaches a preset upper limit number; accordingly, the method for optimizing the initial pressure of the thermal power unit operation further includes:

[0077] When the number of repeated feedback results reaches a preset upper limit, the main steam pressure flow in the iteration round with the smallest deviation from the working parameters in the training stage is selected as the initial operating pressure of the thermal power unit when the training is terminated.

[0078] The training and optimization of the random forest model are explained as follows:

[0079] The heat rate of the thermal power unit is calculated by the energy balance method, that is, heat rate = input heat / output work, and then the correlation function relationship between the thermal power unit operating parameters and the main steam pressure flow is used for linear transformation to obtain the correlation data set between the heat rate and the main steam pressure flow. By combining the function set of the correlation data set between the heat rate and the main steam pressure flow and the correlation data set between the operating parameters and the main steam pressure flow, the correlation data set between the operating parameters and the heat rate is obtained, and it is used as the training data set for optimizing the random forest model.

[0080] The features in the training data set are extracted to obtain a set of feature values, and the set of feature values ​​is used as the best segmentation feature to recursively split the decision tree to obtain multiple decision trees. The specific implementation of this step is the same as the decision tree construction of the existing random forest model.

[0081] The working parameters of the training phase are input into the optimized random forest model, and the main steam pressure flow value with the lowest heat rate in the current state is obtained through the decision tree. The main steam pressure flow value is input into the feedback tree, and the feedback tree performs reverse judgment on the main steam pressure flow value. Each time a main steam pressure flow judgment value corresponding to a node is obtained, that is, the main steam pressure flow value corrected by the feedback tree node weight, the main steam pressure flow judgment value is passed to the comparison logic node in the feedback tree.

[0082] When the main steam pressure flow judgment value reaches the comparison logic node, the comparison logic node compares the main steam pressure flow judgment value with the main steam pressure flow judgment value of other comparison logic nodes topologically connected to it. If it is determined that the comparison result meets the preset comparison logic relationship, the next node in the current feedback tree will be traversed in reverse judgment order, and the main steam pressure flow judgment value corresponding to each node and subsequent steps will be repeatedly executed until all nodes in the current feedback tree are traversed.

[0083] The above-mentioned preset comparison logic relationship can be independently defined according to actual conditions. Since each feedback tree can occupy its own thread for independent calculation, the number of iterations executed by each feedback tree and the number of nodes traversed in the feedback tree are different. The main steam pressure flow judgment value in other feedback trees whose number of iterations is greater than the preset iteration number threshold and whose number of nodes is greater than the preset node number threshold can be used as the benchmark for the current comparison calculation. If the above-mentioned benchmark is not unique, a main steam pressure flow judgment value with the largest number of iterations and the largest number of nodes can also be selected as a benchmark. If the difference between the current main steam pressure flow judgment value and the main steam pressure flow judgment value used as the benchmark is less than the preset difference, it means that the preset comparison logic relationship is met.

[0084] If the above-mentioned reference is not unique, the flow average value of multiple non-unique main steam pressure flow judgment values ​​may also be calculated and the flow average value may be used as the reference.

[0085] In the early stage of iteration, there may be no feedback tree with an iteration number greater than a preset iteration number threshold and a node number greater than a preset node number threshold. In this case, an initial value can be defined. After the iteration number reaches the initial value, the preset iteration number threshold and the preset node number threshold are used as standard parameters for determining the benchmark. Before the initial value is reached, all main steam pressure flow judgment values ​​can be retained to traverse the next node.

[0086] If it is determined that it does not meet the preset comparison logic relationship, the main steam pressure flow judgment value is deleted. Similarly, in the early stage of iteration, an initial value can also be defined. Before the initial value is reached, all main steam pressure flow judgment values ​​can be retained to traverse the next node.

[0087] If the deviation between the feedback result output by the feedback tree and the working parameters in the training phase is less than the set threshold, the main steam pressure flow value obtained by the decision tree is used as the output result of the optimized random forest model.

[0088] If the deviation between the feedback result output by the feedback tree and the working parameters of the training phase is greater than or equal to the set threshold, the input of the decision tree is corrected according to the deviation of the feedback result, and then a prediction is made again until the deviation between the feedback result output by the feedback tree and the working parameters of the training phase is less than the set threshold.

[0089] When the number of repeated feedback results reaches a preset upper limit, the main steam pressure flow in the iteration round with the smallest deviation from the working parameters in the training stage is selected as the initial operating pressure of the thermal power unit when the training ends, that is, as the output result of the optimized random forest model.

[0090] The optimized random forest model has a topological structure of a feedback tree. When performing prediction operations, the output of the decision tree can be cross-validated through the feedback tree, thereby reducing the occurrence of overfitting and improving prediction accuracy.

[0091] The embodiment of the present invention provides a method for optimizing the initial operating pressure of a thermal power unit, which obtains the operating parameters of the thermal power unit in the use phase; processes the operating parameters in the use phase based on an optimized random forest model to obtain the main steam pressure flow when the heat rate reaches the minimum value, and uses the main steam pressure flow at this time as the initial operating pressure of the thermal power unit; wherein the optimized random forest model includes a feedback tree corresponding to each decision tree, and is trained after linear normalization processing is performed on the working parameters in the training phase according to a dynamic vector field; the topological structure of the nodes in each feedback tree is mirror-symmetric to the topological structure of the nodes in the corresponding decision tree, and each feedback tree includes a comparison logic node connected to each node in the feedback tree, and the comparison logic node has a topological connection relationship with the comparison logic nodes corresponding to each other in other feedback trees, which can improve the response speed and control accuracy of the initial operating pressure control of the thermal power unit.

[0092] Furthermore, the linear normalization processing of the working parameters in the training phase according to the dynamic vector field includes:

[0093] The working parameters of the training phase are arranged in chronological order to obtain a working parameter matrix; the above description may be referred to and will not be repeated herein.

[0094] The matrix elements in the working parameter matrix are mapped to the dynamic vector field to obtain a working parameter vector set; the description may refer to the above embodiment and will not be repeated here.

[0095] According to the linear correlation dimension of each working parameter vector in the working parameter vector set, a first-order working parameter dynamic vector value is obtained, and the first-order working parameter dynamic vector value is used as a weight corresponding to each working parameter vector to perform a linear transformation on the working parameter matrix to achieve linear normalization processing. The above-mentioned embodiment can be referred to for explanation and will not be repeated here.

[0096] Further, the obtaining of the first-order working parameter dynamic vector value according to the linear correlation dimension of each working parameter vector in the working parameter vector set includes:

[0097] According to the value of the linear correlation dimension of each working parameter vector in the working parameter vector set, a corresponding single or multiple differential calculation is performed on each working parameter vector to obtain the first-order working parameter dynamic vector value. The above embodiment can be referred to for explanation and will not be repeated here.

[0098] Further, training the optimized random forest model includes:

[0099] According to the training phase operating parameters after linear normalization processing, a correlation data set between the training phase operating parameters and the heat rate is obtained, and the correlation data set is used as the training data set; the description can be made with reference to the above embodiment and will not be repeated here.

[0100] The features in the training data set are extracted to obtain a feature value set, and the feature value set is used as the optimal segmentation feature to recursively split the decision tree to obtain multiple decision trees; the above embodiment can be referred to for description and will not be repeated here.

[0101] Input the working parameters of the training phase to each decision tree, output the main steam pressure flow value corresponding to the minimum heat rate through each decision tree, and input the main steam pressure flow value into the feedback tree corresponding to the decision tree; refer to the above embodiment for description and no further details will be given.

[0102] The feedback result output by each feedback tree is obtained, and the deviation value between each feedback result and the working parameter of the training phase is corrected until the preset iteration termination condition is met. The above-mentioned embodiment can be referred to for explanation and will not be described in detail.

[0103] Further, obtaining the feedback result output by each feedback tree includes:

[0104] The nodes in each feedback tree are traversed respectively to perform reverse judgment on the main steam pressure flow value. Every time a main steam pressure flow judgment value corresponding to a node is obtained, the main steam pressure flow judgment value is transmitted to the comparison logic node in the corresponding feedback tree, and the main steam pressure flow judgment value in the current comparison logic node is compared with the main steam pressure flow judgment values ​​in other comparison logic nodes that have a topological connection relationship with the current comparison logic node. The above-mentioned embodiment can be referred to for description and will not be repeated here.

[0105] If it is determined that the preset comparison logic relationship is met, the next node in the current feedback tree is traversed in reverse judgment order, and the main steam pressure flow judgment value corresponding to each node and subsequent steps are repeatedly executed until all nodes in the current feedback tree are traversed, and the working parameter corresponding to the main steam pressure flow judgment value at this time is used as the feedback result. The above embodiment can be referred to for explanation, and no further description is given.

[0106] Furthermore, the method for optimizing the initial operating pressure of a thermal power unit further comprises:

[0107] If it is determined that the preset comparison logic relationship is not met, the main steam pressure flow determination value is deleted. The above-mentioned embodiment can be referred to for explanation, and no further description is given.

[0108] Furthermore, the preset termination condition includes that the number of repeated feedback results reaches a preset upper limit number; accordingly, the method for optimizing the initial pressure of the thermal power unit operation further includes:

[0109] When the number of repeated feedback results reaches the preset upper limit, the main steam pressure flow in the iteration round with the smallest deviation from the working parameters in the training phase is selected as the initial operating pressure of the thermal power unit when the training ends.

[0110] Figure 2 FIG. 1 is a schematic diagram of a thermal power unit operation initial pressure optimization device provided by an embodiment of the present invention. Figure 2 As shown, the thermal power unit operation initial pressure optimization device provided by the embodiment of the present invention includes an acquisition unit 201 and an optimization unit 202, wherein:

[0111] The acquisition unit 201 is used to acquire the working parameters of the use phase of the thermal power unit; the optimization unit 202 is used to process the working parameters of the use phase based on the optimized random forest model to obtain the main steam pressure flow when the heat rate reaches the minimum value, and use the main steam pressure flow at this time as the initial operating pressure of the thermal power unit; wherein the optimized random forest model includes a feedback tree corresponding to each decision tree, and is trained after linear normalization of the working parameters of the training phase according to the dynamic vector field; the topological structure of the nodes in each feedback tree is mirror-symmetric to the topological structure of the nodes in the corresponding decision tree, and each feedback tree includes a comparison logic node connected to each node in the feedback tree, and the comparison logic node has a topological connection relationship with the corresponding comparison logic nodes in other feedback trees.

[0112] Specifically, the acquisition unit 201 in the device is used to acquire the working parameters of the use phase of the thermal power unit; the optimization unit 202 is used to process the working parameters of the use phase based on the optimized random forest model to obtain the main steam pressure flow when the heat rate reaches the minimum value, and use the main steam pressure flow at this time as the initial operating pressure of the thermal power unit; wherein the optimized random forest model includes a feedback tree corresponding to each decision tree, and is trained after linear normalization of the working parameters in the training phase according to the dynamic vector field; the topological structure of the nodes in each feedback tree is mirror-symmetrical to the topological structure of the nodes in the corresponding decision tree, and each feedback tree respectively includes a comparison logic node connected to each node in the feedback tree, and the comparison logic node has a topological connection relationship with the corresponding comparison logic nodes in other feedback trees.

[0113] The thermal power unit initial pressure optimization device provided in the embodiment of the present invention obtains the working parameters of the thermal power unit in the use phase; processes the working parameters in the use phase based on the optimized random forest model to obtain the main steam pressure flow when the heat rate reaches the minimum value, and uses the main steam pressure flow at this time as the thermal power unit initial pressure; wherein the optimized random forest model includes a feedback tree corresponding to each decision tree, and is trained after linear normalization processing is performed on the working parameters in the training phase according to the dynamic vector field; the topological structure of the nodes in each feedback tree is mirror-symmetrical to the topological structure of the nodes in the corresponding decision tree, and each feedback tree respectively includes a comparison logic node connected to each node in the feedback tree, and the comparison logic node has a topological connection relationship with the comparison logic nodes corresponding to each other in other feedback trees, which can improve the response speed and control accuracy of the thermal power unit initial pressure control.

[0114] The embodiment of the present invention provides an embodiment of the initial pressure optimization device for thermal power unit operation, which can be specifically used to execute the processing flow of the above-mentioned method embodiments. Its functions are not repeated here, and reference can be made to the detailed description of the above-mentioned method embodiments.

[0115] Figure 3 A schematic diagram of the physical structure of an electronic device provided in an embodiment of the present invention, such as Figure 3 As shown, the electronic device includes: a processor (processor) 301, a memory (memory) 302 and a bus 303;

[0116] The processor 301 and the memory 302 communicate with each other via a bus 303;

[0117] The processor 301 is used to call the program instructions in the memory 302 to execute the methods provided by the above method embodiments, for example, including:

[0118] Obtain the working parameters of the thermal power unit during the use phase;

[0119] The working parameters of the use phase are processed based on the optimized random forest model to obtain the main steam pressure flow when the heat rate reaches the minimum value, and the main steam pressure flow at this time is used as the initial operating pressure of the thermal power unit;

[0120] The optimized random forest model includes feedback trees corresponding to each decision tree one by one, and is obtained by training after linear normalization of working parameters in the training phase according to a dynamic vector field; the topological structure of the nodes in each feedback tree is mirror-symmetrical to the topological structure of the nodes in the corresponding decision tree, and each feedback tree includes a comparison logic node connected to each node in the feedback tree, and the comparison logic node has a topological connection relationship with the corresponding comparison logic nodes in other feedback trees.

[0121] This embodiment discloses a computer program product, which includes a computer program stored on a non-transitory computer-readable storage medium. The computer program includes program instructions. When the program instructions are executed by a computer, the computer can perform the methods provided by the above method embodiments, for example, including:

[0122] Obtain the working parameters of the thermal power unit during the use phase;

[0123] The working parameters of the use phase are processed based on the optimized random forest model to obtain the main steam pressure flow when the heat rate reaches the minimum value, and the main steam pressure flow at this time is used as the initial operating pressure of the thermal power unit;

[0124] The optimized random forest model includes feedback trees corresponding to each decision tree one by one, and is obtained by training after linear normalization of working parameters in the training phase according to a dynamic vector field; the topological structure of the nodes in each feedback tree is mirror-symmetrical to the topological structure of the nodes in the corresponding decision tree, and each feedback tree includes a comparison logic node connected to each node in the feedback tree, and the comparison logic node has a topological connection relationship with the corresponding comparison logic nodes in other feedback trees.

[0125] This embodiment provides a computer-readable storage medium, wherein the computer-readable storage medium stores a computer program, wherein the computer program enables the computer to execute the methods provided by the above method embodiments, for example, including:

[0126] Obtain the working parameters of the thermal power unit during the use phase;

[0127] The working parameters of the use phase are processed based on the optimized random forest model to obtain the main steam pressure flow when the heat rate reaches the minimum value, and the main steam pressure flow at this time is used as the initial operating pressure of the thermal power unit;

[0128] The optimized random forest model includes feedback trees corresponding to each decision tree one by one, and is obtained by training after linear normalization of working parameters in the training phase according to a dynamic vector field; the topological structure of the nodes in each feedback tree is mirror-symmetrical to the topological structure of the nodes in the corresponding decision tree, and each feedback tree includes a comparison logic node connected to each node in the feedback tree, and the comparison logic node has a topological connection relationship with the corresponding comparison logic nodes in other feedback trees.

[0129] Those skilled in the art will appreciate that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0130] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to produce a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0131] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0132] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1 A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0133] In the description of this specification, the description with reference to the terms "one embodiment", "a specific embodiment", "some embodiments", "for example", "example", "specific example", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0134] The specific embodiments described above further illustrate the objectives, technical solutions and beneficial effects of the present invention in detail. It should be understood that the above description is only a specific embodiment of the present invention and is not intended to limit the scope of protection of the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the scope of protection of the present invention.

Claims

1. A method for optimizing the initial pressure of a thermal power unit, characterized in that: include: Obtain the working parameters of the thermal power unit during the use phase; The working parameters of the use phase are processed based on the optimized random forest model to obtain the main steam pressure flow when the heat rate reaches the minimum value, and the main steam pressure flow at this time is used as the initial operating pressure of the thermal power unit; The optimized random forest model includes feedback trees corresponding to each decision tree one by one, and is obtained by training after linear normalization of working parameters in the training phase according to a dynamic vector field; the topological structure of the nodes in each feedback tree is mirror-symmetrical to the topological structure of the nodes in the corresponding decision tree, and each feedback tree includes a comparison logic node connected to each node in the feedback tree, and the comparison logic node has a topological connection relationship with the corresponding comparison logic nodes in other feedback trees.

2. The method for optimizing the initial operating pressure of a thermal power unit according to claim 1, characterized in that: The linear normalization processing of the working parameters in the training phase according to the dynamic vector field includes: Arranging the working parameters of the training phase in chronological order to obtain a working parameter matrix; Mapping the matrix elements in the working parameter matrix to the dynamic vector field to obtain a working parameter vector set; According to the linear correlation dimension of each working parameter vector in the working parameter vector set, a first-order working parameter dynamic vector value is obtained, and the first-order working parameter dynamic vector value is used as the weight corresponding to each working parameter vector to perform linear transformation on the working parameter matrix to achieve linear normalization processing.

3. The method for optimizing the initial operating pressure of a thermal power unit according to claim 2, characterized in that: The obtaining of the first-order working parameter dynamic vector value according to the linear correlation dimension of each working parameter vector in the working parameter vector set includes: According to the value of the linear correlation dimension of each working parameter vector in the working parameter vector set, a corresponding single or multiple differential calculation is performed on each working parameter vector to obtain the first-order working parameter dynamic vector value.

4. The method for optimizing the initial operating pressure of a thermal power unit according to claim 1, characterized in that: Training the optimized random forest model includes: According to the training phase working parameters after linear normalization processing, a correlation data set between the training phase working parameters and the heat rate is obtained, and the correlation data set is used as the training data set; Extracting features from the training data set to obtain a feature value set, and using the feature value set as the best segmentation feature to recursively split the decision tree to obtain multiple decision trees; Inputting the working parameters of the training phase into each decision tree, outputting the main steam pressure flow value corresponding to the minimum heat rate through each decision tree, and inputting the main steam pressure flow value into the feedback tree corresponding to the decision tree; The feedback result output by each feedback tree is obtained, and the deviation value between each feedback result and the working parameter of the training stage is corrected until a preset iteration termination condition is met.

5. The method for optimizing the initial pressure of a thermal power unit according to claim 4, characterized in that: The obtaining of the feedback result output by each feedback tree includes: Traversing the nodes in each feedback tree respectively to perform reverse judgment on the main steam pressure flow value, and when each main steam pressure flow judgment value corresponding to a node is obtained, the main steam pressure flow judgment value is transmitted to the comparison logic node in the corresponding feedback tree, and the main steam pressure flow judgment value in the current comparison logic node is compared with the main steam pressure flow judgment values ​​in other comparison logic nodes that have a topological connection relationship with the current comparison logic node; If it is determined that the preset comparison logic relationship is met, continue to traverse the next node in the current feedback tree in the reverse judgment order, and repeat the main steam pressure flow judgment value corresponding to each node and subsequent steps until all nodes in the current feedback tree are traversed, and the working parameters corresponding to the main steam pressure flow judgment value at this time are used as the feedback results.

6. The method for optimizing the initial pressure of a thermal power unit according to claim 4, characterized in that: The method for optimizing the initial operating pressure of a thermal power unit further comprises: If it is determined that the preset comparison logic relationship is not met, the main steam pressure flow determination value is deleted.

7. The method for optimizing the initial pressure of a thermal power unit according to claim 4, characterized in that: The preset termination condition includes that the number of repeated feedback results reaches a preset upper limit number; accordingly, the method for optimizing the initial pressure of the thermal power unit operation further includes: When the number of repeated feedback results reaches a preset upper limit, the main steam pressure flow in the iteration round with the smallest deviation from the working parameters in the training stage is selected as the initial operating pressure of the thermal power unit when the training is terminated.

8. A device for optimizing the initial pressure of a thermal power unit, characterized in that: include: An acquisition unit, used for acquiring the working parameters of the thermal power unit during the use phase; An optimization unit is used to process the working parameters of the use phase based on an optimization random forest model to obtain the main steam pressure flow when the heat rate reaches a minimum value, and use the main steam pressure flow at this time as the initial operating pressure of the thermal power unit; The optimized random forest model includes feedback trees corresponding to each decision tree one by one, and is obtained by training after linear normalization of working parameters in the training phase according to a dynamic vector field; the topological structure of the nodes in each feedback tree is mirror-symmetrical to the topological structure of the nodes in the corresponding decision tree, and each feedback tree includes a comparison logic node connected to each node in the feedback tree, and the comparison logic node has a topological connection relationship with the corresponding comparison logic nodes in other feedback trees.

9. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method according to any one of claims 1 to 7 are implemented.

Images