A steam pressure and flow automatic control system and control method thereof

Abstract

The present invention discloses an automatic steam pressure and flow control system, comprising a steam pressure detection module for detecting steam pressure data; a steam flow detection module for detecting steam flow data; a unit heat consumption calculation module for calculating real-time unit heat consumption values; a detection data preprocessing module for preprocessing steam pressure and steam flow data; and a unit heat consumption optimization module for optimizing unit heat consumption values by adjusting steam pressure and steam flow data. The present invention can overcome the shortcomings of existing technologies and improve the combined control effect of steam pressure and flow in thermal power units.

Description

Technical Field

[0001] The present invention relates to the technical field of thermal power unit control, in particular to a steam pressure and flow automatic control system and a control method thereof. Background Art

[0002] my country is a major powerhouse in thermal power generation, and improving the operating efficiency of thermal power units has long been a research hotspot. Among the many operating parameters of thermal power units, steam pressure and flow rate are two of the most significant influencing factors. However, due to their inherent coupling, controlling these two parameters independently is ineffective. Summary of the Invention

[0003] The technical problem to be solved by the present invention is to provide a steam pressure and flow automatic control system and a control method thereof, which can solve the deficiencies of the existing technology and improve the joint control effect of steam pressure and flow in thermal power units.

[0004] In order to solve the above technical problems, the technical solutions adopted by the present invention are as follows.

[0005] A steam pressure and flow automatic control system, comprising:

[0006] Steam pressure detection module, used to detect steam pressure data;

[0007] Steam flow detection module, used to detect steam flow data;

[0008] Unit heat consumption value calculation module, used to calculate the real-time unit heat consumption value;

[0009] Detection data preprocessing module, used for preprocessing steam pressure data and steam flow data;

[0010] The unit heat consumption value optimization module optimizes the unit heat consumption value by adjusting the steam pressure data and steam flow data.

[0011] A control method for the above-mentioned steam pressure and flow automatic control system comprises the following steps:

[0012] A. The steam pressure detection module and the steam flow detection module detect the steam pressure data and steam flow data respectively. The unit heat consumption value calculation module calculates the real-time unit heat consumption value according to the actual operating status of the unit;

[0013] B. The detection data preprocessing module performs curve fitting on the discrete steam pressure data, steam flow data, and unit heat consumption value. It then compares the steam pressure curve and steam flow curve with the unit heat consumption value curve, and marks the high-weight data corresponding to the current area on the unit heat consumption value curve.

[0014] C. The unit heat consumption value optimization module establishes a neural network model and uses historical data to train the neural network model. Then, the current steam pressure data and steam flow data as well as other unit parameter constraints are input to obtain the optimized steam pressure data and steam flow data.

[0015] Preferably, in step B, marking the high-weight data corresponding to the current area on the unit heat consumption value curve includes the following steps:

[0016] B1. Preset a sliding window and use it to synchronously slide and obtain values on the steam pressure curve, steam flow curve, and unit heat consumption curve;

[0017] B2. Establish mapping relationships between the steam pressure curve segment and the steam flow curve segment within the sliding window and the unit heat consumption curve segment;

[0018] B3. Fit the obtained mapping relationship to a Gaussian distribution based on linear similarity, and take the mapping relationship at the midpoint of the Gaussian distribution as the reference mapping relationship;

[0019] B4. Compare the mapping relationship corresponding to the steam pressure curve segment and the steam flow curve segment in the sliding window with the reference mapping relationship. If the linear similarity between the mapping relationship of the steam pressure curve segment and the steam flow curve segment and the reference mapping relationship is greater than the set threshold, then take the data corresponding to the mapping relationship with the larger linear similarity as the high-weight data. If the linear similarity between the mapping relationship of the steam pressure curve segment or the steam flow curve segment and the reference mapping relationship is greater than the set threshold, then take the data corresponding to the mapping relationship as the high-weight data. If the linear similarity between the mapping relationship of the steam pressure curve segment and the steam flow curve segment and the reference mapping relationship is less than or equal to the set threshold, return to step B1, reduce the width of the sliding window and recalculate until all unit heat consumption value curves are marked and end.

[0020] Preferably, in step C, the neural network model includes an output layer, three hidden layers and an output layer.

[0021] Preferably, in step C, when training the neural network model, the neural network model is first trained using high-weight data to determine the adjustment range of the neural network model parameters, and then the neural network model is trained using non-high-weight data, and the neural network model parameters are adjusted a second time within the adjustment range of the neural network model parameters.

[0022] Preferably, in step C, the unit is controlled using the optimized steam pressure data and steam flow data, and the set threshold in step B is adjusted according to the deviation between the actual heat consumption value and the theoretical heat consumption value of the unit.

[0023] The beneficial effect brought about by the adoption of the above technical solution is that the present invention adopts a neural network model to optimize the control parameters. Since the loss function of the neural network model optimized for coupled parameters fluctuates significantly around the stable value during the training process, the training instability is relatively large. In response to this problem, the present invention specially pre-processes the steam pressure data and steam flow data, and sets high-weight data in segments according to its association with the unit heat consumption value. Then, during the neural network training process, the two parameters to be controlled are decoupled and two-level training is performed, which effectively reduces the instability during the training process and improves the training speed. In addition, by feedback adjustment of the set threshold, the data obtained during the use of the neural network model can be used to optimize its future retraining process. BRIEF DESCRIPTION OF THE DRAWINGS

[0024] Figure 1 It is a schematic diagram of a specific embodiment of the present invention.

[0025] In the figure: 1. Steam pressure detection module; 2. Steam flow detection module; 3. Unit heat consumption value calculation module; 4. Detection data preprocessing module; 5. Unit heat consumption value optimization module. DETAILED DESCRIPTION

[0026] Reference Figure 1 , a specific embodiment of the present invention includes,

[0027] Steam pressure detection module 1, used for detecting steam pressure data;

[0028] Steam flow detection module 2, used for detecting steam flow data;

[0029] Unit heat consumption value calculation module 3, used to calculate the real-time unit heat consumption value;

[0030] Detection data preprocessing module 4, used for preprocessing steam pressure data and steam flow data;

[0031] The unit heat consumption value optimization module 5 optimizes the unit heat consumption value by adjusting the steam pressure data and the steam flow data.

[0032] A control method for the above-mentioned steam pressure and flow automatic control system comprises the following steps:

[0033] A. The steam pressure detection module 1 and the steam flow detection module 2 detect the steam pressure data and the steam flow data respectively, and the unit heat consumption value calculation module 3 calculates the real-time unit heat consumption value according to the actual operating status of the unit;

[0034] B. The detection data preprocessing module 4 performs curve fitting on the discrete steam pressure data, steam flow data, and unit heat consumption value, and then compares the steam pressure curve and steam flow curve with the unit heat consumption value curve, marking the high-weight data corresponding to the current area on the unit heat consumption value curve;

[0035] C. The unit heat consumption value optimization module 5 establishes a neural network model and uses historical data to train the neural network model. Then, the current steam pressure data and steam flow data as well as other unit parameter constraints are input to obtain the optimized steam pressure data and steam flow data.

[0036] In step B, marking the high-weight data corresponding to the current area on the unit heat consumption value curve includes the following steps:

[0037] B1. Preset a sliding window and use it to synchronously slide and obtain values on the steam pressure curve, steam flow curve, and unit heat consumption curve;

[0038] B2. Establish mapping relationships between the steam pressure curve segment and the steam flow curve segment within the sliding window and the unit heat consumption curve segment;

[0039] B3. Fit the obtained mapping relationship to a Gaussian distribution based on linear similarity, and take the mapping relationship at the midpoint of the Gaussian distribution as the reference mapping relationship;

[0040] B4. Compare the mapping relationship corresponding to the steam pressure curve segment and the steam flow curve segment in the sliding window with the reference mapping relationship. If the linear similarity between the mapping relationship of the steam pressure curve segment and the steam flow curve segment and the reference mapping relationship is greater than the set threshold, then take the data corresponding to the mapping relationship with the larger linear similarity as the high-weight data. If the linear similarity between the mapping relationship of the steam pressure curve segment or the steam flow curve segment and the reference mapping relationship is greater than the set threshold, then take the data corresponding to the mapping relationship as the high-weight data. If the linear similarity between the mapping relationship of the steam pressure curve segment and the steam flow curve segment and the reference mapping relationship is less than or equal to the set threshold, return to step B1, reduce the width of the sliding window and recalculate until all unit heat consumption value curves are marked and end.

[0041] In step C, the neural network model includes an output layer, three hidden layers and an output layer.

[0042] In step C, when training the neural network model, the neural network model is first trained using high-weight data to determine the adjustment range of the neural network model parameters, and then the neural network model is trained using non-high-weight data, and the neural network model parameters are adjusted a second time within the adjustment range of the neural network model parameters.

[0043] In step C, the optimized steam pressure data and steam flow data are used to control the unit, and the set threshold in step B is adjusted according to the deviation between the actual heat consumption value and the theoretical heat consumption value of the unit.

[0044] The present invention effectively improves the combined control effect of steam pressure and flow by establishing a neural network model and optimizing its training process.

[0045] The basic principles, main features, and advantages of the present invention are shown and described above. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The above embodiments and descriptions are merely illustrative of the principles of the present invention. Various changes and modifications may be made to the present invention without departing from the spirit and scope of the present invention. Such changes and modifications are intended to fall within the scope of the present invention. The scope of protection claimed in the present invention is defined by the appended claims and their equivalents.

Claims

1. A control method for a steam pressure and flow automatic control system, characterized in that The following steps are involved: A. The steam pressure detection module (1) and the steam flow detection module (2) detect steam pressure data and steam flow data respectively, and the unit heat consumption value calculation module (3) calculates the real-time unit heat consumption value according to the actual operating status of the unit; B. The detection data preprocessing module (4) performs curve fitting on the discrete steam pressure data, steam flow data and unit heat consumption value respectively, and then uses the steam pressure curve and steam flow curve to compare with the unit heat consumption value curve respectively, and marks the high-weight data corresponding to the current area on the unit heat consumption value curve; marking the high-weight data corresponding to the current area on the unit heat consumption value curve includes the following steps: B1. Preset a sliding window and use it to synchronously slide and obtain values on the steam pressure curve, steam flow curve, and unit heat consumption curve; B2. Establish mapping relationships between the steam pressure curve segment and the steam flow curve segment within the sliding window and the unit heat consumption curve segment; B3. Fit the obtained mapping relationship to a Gaussian distribution based on linear similarity, and take the mapping relationship at the midpoint of the Gaussian distribution as the reference mapping relationship; B4. Compare the mapping relationships corresponding to the steam pressure curve segments and steam flow curve segments within the sliding window with the reference mapping relationships. If the linear similarities between the mapping relationships of the steam pressure curve segments and steam flow curve segments and the reference mapping relationships are both greater than a set threshold, then take the data corresponding to the mapping relationships with the greater linear similarity as high-weight data. If the linear similarities between the mapping relationships of the steam pressure curve segments or steam flow curve segments and the reference mapping relationships are greater than a set threshold, then take the data corresponding to the mapping relationships as high-weight data. If the linear similarities between the mapping relationships of the steam pressure curve segments and steam flow curve segments and the reference mapping relationships are both less than or equal to the set threshold, then return to step B1, reduce the width of the sliding window, and recalculate until all unit heat consumption curves are marked and the process ends. C. Unit heat consumption value optimization module (5) establishes a neural network model and uses historical data to train the neural network model, then inputs the current steam pressure data and steam flow data and other unit parameter constraints to obtain optimized steam pressure data and steam flow data; when training the neural network model, first use high-weight data to train the neural network model to determine the adjustment range of the neural network model parameters, then use non-high-weight data to train the neural network model, and perform secondary adjustment on the neural network model parameters within the adjustment range of the neural network model parameters.

2. The control method of the steam pressure and flow automatic control system according to claim 1, characterized in that: In step C, the neural network model includes an output layer, three hidden layers and an output layer.

3. The control method of the steam pressure and flow automatic control system according to claim 2, characterized in that: In step C, the optimized steam pressure data and steam flow data are used to control the unit, and the set threshold in step B is adjusted according to the deviation between the actual heat consumption value and the theoretical heat consumption value of the unit.

Images