Intelligent steam supply method for thermal power plant

Abstract

The invention discloses an intelligent steam supply method for a thermal power plant, and relates to the technical field of thermal power generation automatic control, and the method comprises the steps: building a steam supply training set according to historical steam supply parameters, inputting the steam supply training set into a preset machine learning model, and carrying out the training, thereby obtaining a steam supply prediction model; the steam supply prediction model receives real-time steam supply parameters of the thermal power plant and outputs target steam supply parameters based on the real-time steam supply parameters; the target steam supply parameter controls the opening degree of a pipe network valve through an execution unit, so that the actual steam supply parameter of the steam supply pipe network of the thermal power plant is matched with the target steam supply parameter. Through the synergistic effect of the LSTM prediction model and the fuzzy PID controller, intelligent control over the thermal power plant steam supply system is achieved, the system safety is guaranteed, meanwhile, the prediction precision and control performance of steam supply parameters are remarkably improved, equipment abrasion is reduced, and the system operation efficiency is improved.

Description

Technical Field

[0001] The invention relates to the technical field of thermal power generation automatic control, and in particular to an intelligent steam supply method for a thermal power plant. Background Art

[0002] In the field of thermal power generation, the steam supply system plays a core role in industrial production. Traditional thermal power plant steam supply systems mostly adopt a constant pressure and quantitative steam supply method, and the PID controller is used to adjust the pipe network valve opening to maintain the pipe network pressure. This control method has obvious limitations: First, the parameter setting process of the PID controller is cumbersome, requiring technicians to repeatedly debug based on experience, and once the parameters are determined, it is difficult to adapt to changes in operating conditions; second, due to the drastic dynamic changes in steam load, the pipe network system has significant lag characteristics, resulting in slow system response and unable to meet the needs of rapid adjustment; third, the multivariable coupling characteristics of the steam supply system are significant, and there are complex nonlinear relationships between the parameters. Traditional PID control strategies are difficult to achieve precise control of the system, resulting in energy waste.

[0003] In recent years, data-driven steam supply control methods have gradually attracted attention. Some researchers have proposed using intelligent algorithms such as neural networks to build system models and obtain prediction models through training with historical operation data. However, there are still many problems with existing intelligent steam supply methods: first, the dynamic characteristics of the steam supply system are not fully considered during the model training process, resulting in insufficient prediction accuracy; second, the prediction results are disconnected from the actual execution link and lack an effective closed-loop control mechanism; third, existing methods generally ignore the safety constraints during system operation and pose potential operational risks. These problems have seriously restricted the engineering application of intelligent steam supply technology. Summary of the invention

[0004] In view of the above-mentioned problems, the present invention is proposed.

[0005] Therefore, the present invention provides an intelligent steam supply method for a thermal power plant, which can solve the problems mentioned in the background technology.

[0006] In order to solve the above technical problems, the present invention provides the following technical solutions: an intelligent steam supply method for a thermal power plant, comprising: establishing a steam supply training set according to historical steam supply parameters, inputting the steam supply training set into a preset machine learning model for training to obtain a steam supply prediction model;

[0007] The steam supply prediction model receives the real-time steam supply parameters of the thermal power plant and outputs the target steam supply parameters based on the real-time steam supply parameters;

[0008] The target steam supply parameter controls the opening of the pipe network valve via the execution unit, so that the actual steam supply parameter of the steam supply pipe network of the thermal power plant matches the target steam supply parameter.

[0009] As a preferred solution of the intelligent steam supply method for a thermal power plant described in the present invention, a steam supply training set is established according to historical steam supply parameters, and the steam supply training set is input into a preset machine learning model for training to obtain a steam supply prediction model, including the following steps:

[0010] Collect the following historical steam supply parameters: steam supply pressure, steam supply temperature, pipe network valve opening, steam supply flow rate and heat load demand of the steam supply network;

[0011] Dividing the historical steam supply parameters into time series according to the collection time;

[0012] The historical steam supply parameters corresponding to each time point are constructed as a feature vector, and the steam supply pressure, steam supply temperature and steam supply flow rate at the next time point are constructed as a label vector;

[0013] If the steam supply pressure in the feature vector is less than a preset pressure threshold or the steam supply temperature is less than a preset temperature threshold, the feature vector and the corresponding label vector are marked as abnormal samples;

[0014] Combining the feature vectors and label vectors that are not marked as abnormal samples to form a steam supply training set;

[0015] The steam supply training set is input into a preset long short-term memory network LSTM model for training to obtain a steam supply prediction model.

[0016] As a preferred solution of the intelligent steam supply method for a thermal power plant described in the present invention, the steam supply training set is input into a preset long short-term memory network LSTM model for training to obtain a steam supply prediction model, including the following steps:

[0017] Dividing the steam supply training set into a training data set, a validation data set and a test data set in a ratio of 8:1:1;

[0018] Construct a neural network structure with three layers of LSTM units, where the number of neurons in each layer of LSTM units is 128, 64, and 32 respectively;

[0019] If the number of training rounds is less than 1000 and the validation set loss value has not decreased for 5 consecutive rounds, the learning rate is reduced to 0.8 times the original value;

[0020] If the validation set loss value does not decrease for 10 consecutive rounds, the training process is stopped and the current model parameters are saved;

[0021] Use the test data set to test the trained model and output the prediction accuracy.

[0022] As a preferred solution of the intelligent steam supply method for a thermal power plant described in the present invention, the steam supply prediction model receives the real-time steam supply parameters of the thermal power plant, and outputs the target steam supply parameters based on the real-time steam supply parameters, including the following steps:

[0023] Real-time collection of real-time steam supply parameters in the steam supply system of thermal power plants, including: real-time steam supply pressure, real-time steam supply temperature, real-time pipe network valve opening, real-time steam supply flow and real-time heat load demand;

[0024] The real-time steam supply parameters are constructed into a feature vector in the same format as the step of "constructing the historical steam supply parameters corresponding to each time sequence point into a feature vector, and constructing the steam supply pressure, steam supply temperature and steam supply flow rate at the next time sequence point into a label vector";

[0025] Inputting the feature vector into the trained steam supply prediction model;

[0026] The steam supply prediction model outputs the predicted steam supply pressure, predicted steam supply temperature and predicted steam supply flow rate at the next time point, and these three parameters are used as target steam supply parameters;

[0027] If the predicted steam supply pressure exceeds the safety pressure range or the predicted steam supply temperature exceeds the safety temperature range, the target steam supply parameter is set to a boundary value of the corresponding parameter range;

[0028] The target steam supply parameters, real-time steam supply parameters and corresponding time information are stored in a historical database for subsequent online updating of the model.

[0029] As a preferred solution of the intelligent steam supply method for a thermal power plant described in the present invention, wherein: setting the target steam supply parameter to the boundary value of the corresponding parameter range comprises the following steps: setting the target steam supply parameter to the boundary value of the corresponding parameter range comprises:

[0030] The safety range of steam supply pressure is set to [Pmin, Pmax], where Pmin is the minimum safety pressure value and Pmax is the maximum safety pressure value; the safety range of steam supply temperature is set to [Tmin, Tmax], where Tmin is the minimum safety temperature value and Tmax is the maximum safety temperature value;

[0031] If the predicted steam supply pressure is less than Pmin, the target steam supply pressure is set to Pmin+ΔP, where ΔP is the pressure safety margin value;

[0032] If the predicted steam supply pressure is greater than Pmax, the target steam supply pressure is set to Pmax-ΔP;

[0033] If the predicted steam supply temperature is less than Tmin, the target steam supply temperature is set to Tmin+ΔT, where ΔT is the temperature safety margin value;

[0034] If the predicted steam supply temperature is greater than Tmax, the target steam supply temperature is set to Tmax-ΔT;

[0035] Calculate the deviation between the target steam supply parameters and the predicted steam supply parameters. If the deviation exceeds the preset threshold, an alarm signal is triggered and the alarm information is recorded.

[0036] As a preferred solution of the intelligent steam supply method for a thermal power plant of the present invention, wherein: the target steam supply parameter controls the opening of the pipe network valve via the execution unit, so that the actual steam supply parameter of the steam supply pipe network of the thermal power plant matches the target steam supply parameter, including the following steps:

[0037] Comparing the target steam supply parameter with the real-time steam supply parameter, and calculating the steam supply pressure deviation value, the steam supply temperature deviation value and the steam supply flow deviation value;

[0038] Based on the deviation value, a fuzzy PID controller is used to calculate the target opening of each pipe network valve;

[0039] If the calculated target opening change rate is greater than the preset opening change rate threshold, the target opening is adjusted step by step according to the preset opening change rate threshold;

[0040] The execution unit receives the target opening and controls each pipe network valve to adjust the opening;

[0041] Collect the actual steam supply parameters after adjustment and calculate the matching degree with the target steam supply parameters;

[0042] If the matching degree is less than the preset matching threshold and still fails to reach the preset matching threshold after three consecutive adjustments, an abnormal processing signal is triggered.

[0043] As a preferred solution of the intelligent steam supply method for a thermal power plant described in the present invention, the target opening of each pipe network valve is calculated using a fuzzy PID controller based on the deviation value, including the following steps:

[0044] A fuzzy rule set is established based on the deviation value, and the steam supply pressure deviation, steam supply temperature deviation and steam supply flow deviation are fuzzified into five levels of {negative large, negative small, zero, positive small, positive large};

[0045] Design a fuzzy rule base to determine the adjustment coefficients of PID parameters according to the deviation level combination, where the adjustment coefficient range of Kp is [0.5, 1.5], the adjustment coefficient range of Ki is [0.3, 1.2], and the adjustment coefficient range of Kd is [0.4, 1.3];

[0046] If the steam supply pressure deviation is at the "negative large" level and the steam supply flow deviation is at the "positive large" level, the Kp adjustment coefficient is set to 1.5 and the Ki adjustment coefficient is set to 0.3;

[0047] The centroid method is used to defuzzify the fuzzy reasoning results and obtain the real-time parameter values ​​of the PID controller;

[0048] Calculate the incremental PID controller output according to the real-time parameter value;

[0049] The output is converted into a target opening value of each pipe network valve. If the change of the target opening value exceeds 2%, it is gradually adjusted in steps of 2%.

[0050] To further solve the above technical problems, the present invention provides the following technical solutions: an intelligent steam supply system for a thermal power plant, comprising: a training module, used to establish a steam supply training set according to historical steam supply parameters, and input the steam supply training set into a preset machine learning model for training to obtain a steam supply prediction model; an output module, the steam supply prediction model receives the real-time steam supply parameters of the thermal power plant, and outputs target steam supply parameters based on the real-time steam supply parameters; a control module, the target steam supply parameters control the opening of the pipeline valve via the execution unit, so that the actual steam supply parameters of the thermal power plant steam supply pipeline network match the target steam supply parameters.

[0051] A computer device comprises a memory and a processor, wherein the memory stores a computer program, and wherein the processor implements the steps of the above-mentioned intelligent steam supply method for a thermal power plant when executing the computer program.

[0052] 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 intelligent steam supply method for a thermal power plant as described above are implemented.

[0053] Beneficial effects of the present invention: The present invention establishes a steam supply training set, uses a machine learning model to train historical data, constructs an accurate steam supply prediction model, and links the prediction results with the execution unit to achieve precise adjustment of the pipe network valve opening. The present invention has the following beneficial effects: first, it improves the prediction accuracy and control accuracy of the steam supply system; second, it achieves seamless connection between prediction and control to ensure stable operation of the system; third, it reduces system energy consumption and improves economic benefits. BRIEF DESCRIPTION OF THE DRAWINGS

[0054] In order to more clearly illustrate the technical solutions of the embodiments of the present invention, the accompanying drawings required for use in the description of the embodiments will be briefly introduced below. Obviously, the accompanying drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other accompanying drawings can be obtained based on these accompanying drawings without paying creative work.

[0055] Figure 1 This is a schematic diagram of the overall process of an intelligent steam supply method for a thermal power plant proposed by the present invention;

[0056] Figure 2 This is a diagram of computer equipment in an intelligent steam supply method for a thermal power plant proposed by the present invention. DETAILED DESCRIPTION

[0057] In order to make the above-mentioned purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, but not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary persons in the art without creative work should fall within the scope of protection of the present invention.

[0058] In the following description, many specific details are set forth to facilitate a full understanding of the present invention, but the present invention may also be implemented in other ways different from those described herein, and those skilled in the art may make similar generalizations without violating the connotation of the present invention. Therefore, the present invention is not limited to the specific embodiments disclosed below.

[0059] Example 1, reference Figure 1 , which is an embodiment of the present invention, provides an intelligent steam supply method for a thermal power plant.

[0060] S1: A steam supply training set is established according to historical steam supply parameters, and the steam supply training set is input into a preset machine learning model for training to obtain a steam supply prediction model.

[0061] S1.1: Collect the following historical steam supply parameters: steam supply pressure, steam supply temperature, pipe network valve opening, steam supply flow rate and heat load demand of the steam supply network;

[0062] S1.2: The historical steam supply parameters are divided into time series according to the collection time, where the time interval between two adjacent time series points is 10 minutes;

[0063] S1.3: construct the historical steam supply parameters corresponding to each time point into a feature vector, and construct the steam supply pressure, steam supply temperature and steam supply flow rate of the next time point into a label vector;

[0064] S1.4: If the steam supply pressure in the feature vector is less than the preset pressure threshold or the steam supply temperature is less than the preset temperature threshold, the feature vector and the corresponding label vector are marked as abnormal samples;

[0065] S1.5: Combine the feature vectors and label vectors that are not marked as abnormal samples to form a steam supply training set;

[0066] S1.6: The steam supply training set is input into a preset long short-term memory network LSTM model for training to obtain a steam supply prediction model.

[0067] Specifically, the steam supply training set is divided into a training data set, a validation data set and a test data set in a ratio of 8:1:1;

[0068] Construct a neural network structure with three layers of LSTM units, where the number of neurons in each layer of LSTM units is 128, 64, and 32 respectively;

[0069] If the number of training rounds is less than 1000 and the validation set loss value has not decreased for 5 consecutive rounds, the learning rate is reduced to 0.8 times the original value;

[0070] If the validation set loss value does not decrease for 10 consecutive rounds, the training process is stopped and the current model parameters are saved;

[0071] Use the test data set to test the trained model and output the prediction accuracy.

[0072] It should be noted that the present invention uses the LSTM model to process the time series data of the steam supply system, which has a special technical effect. Although the traditional feedforward neural network can also process time series data, it cannot effectively capture the long-term dependencies in the steam supply system. The LSTM model can simultaneously process short-term and long-term time series dependencies by introducing memory units and gating mechanisms. Specifically: the memory unit saves the influence of historical steam supply parameters, the forgetting gate filters the importance of historical information, the input gate determines the amount of information update at the current moment, and the output gate controls the final predicted output. This structure is particularly suitable for processing the lag effect and nonlinear characteristics in the steam supply system.

[0073] During the model training process, the present invention adopts dynamic learning rate adjustment and early stopping mechanism. When the validation set loss value does not decrease for 5 consecutive rounds, the model parameters are optimized by reducing the learning rate. This strategy can help the model jump out of the local optimal solution. At the same time, the validation set loss value does not decrease for 10 consecutive rounds as an early stopping condition, which can effectively prevent the model from overfitting. After experimental verification, this training strategy can reduce the prediction error of steam supply parameters by about 15% compared with the method of fixed learning rate and fixed training rounds.

[0074] In addition, the three-layer LSTM structure of the present invention has unique advantages: the 128 neurons in the first layer mainly extract the basic features of the steam supply parameters, the 64 neurons in the second layer perform feature combination, and the 32 neurons in the third layer complete the final feature mapping. This decreasing network structure effectively reduces the complexity of the model, while ensuring the prediction accuracy, and shortens the model training time by about 30%. Through the combined application of the above-mentioned technical means, the present invention solves the technical problems of insufficient accuracy and low training efficiency of the prediction model of the steam supply system.

[0075] S2: The steam supply prediction model receives the real-time steam supply parameters of the thermal power plant and outputs the target steam supply parameters based on the real-time steam supply parameters.

[0076] S2.1: Real-time collection of steam supply parameters in the steam supply system of thermal power plants, including: real-time steam supply pressure, real-time steam supply temperature, real-time pipe network valve opening, real-time steam supply flow rate and real-time heat load demand;

[0077] S2.2: construct the real-time steam supply parameters into a feature vector in the same format as the step "construct the historical steam supply parameters corresponding to each time sequence point into a feature vector, and construct the steam supply pressure, steam supply temperature and steam supply flow rate at the next time sequence point into a label vector";

[0078] S2.3: inputting the feature vector into the trained steam supply prediction model;

[0079] S2.4: The steam supply prediction model outputs the predicted steam supply pressure, predicted steam supply temperature and predicted steam supply flow rate at the next time point, and these three parameters are used as target steam supply parameters;

[0080] S2.5: If the predicted steam supply pressure exceeds the safe pressure range or the predicted steam supply temperature exceeds the safe temperature range, the target steam supply parameter is set to the boundary value of the corresponding parameter range:

[0081] The safety range of steam supply pressure is set to [Pmin, Pmax], where Pmin is the minimum safety pressure value and Pmax is the maximum safety pressure value; the safety range of steam supply temperature is set to [Tmin, Tmax], where Tmin is the minimum safety temperature value and Tmax is the maximum safety temperature value;

[0082] If the predicted steam supply pressure is less than Pmin, the target steam supply pressure is set to Pmin+ΔP, where ΔP is the pressure safety margin value;

[0083] If the predicted steam supply pressure is greater than Pmax, the target steam supply pressure is set to Pmax-ΔP;

[0084] If the predicted steam supply temperature is less than Tmin, the target steam supply temperature is set to Tmin+ΔT, where ΔT is the temperature safety margin value;

[0085] If the predicted steam supply temperature is greater than Tmax, the target steam supply temperature is set to Tmax-ΔT;

[0086] Calculate the deviation between the target steam supply parameters and the predicted steam supply parameters. If the deviation exceeds the preset threshold, an alarm signal is triggered and the alarm information is recorded.

[0087] S2.6: Store the target steam supply parameters, real-time steam supply parameters and corresponding time information into the historical database for subsequent online updating of the model.

[0088] The real-time steam supply parameters include five key indicators: real-time steam supply pressure, real-time steam supply temperature, real-time pipe network valve opening, real-time steam supply flow rate and real-time heat load demand. These five parameters constitute a complete description of the steam supply system status.

[0089] Step S2.2 uses the same feature vector format as S1.3, which has a special technical effect. In traditional methods, real-time data and historical data are often processed in different ways, which easily leads to inconsistent features. The present invention ensures that the input data of the prediction model has the same statistical characteristics as the training data by unifying the feature format. Experimental data show that this processing method reduces the prediction error by about 20%. Specifically, when the steam supply pressure fluctuates in the range of 4MPa to 6MPa, the prediction deviation is controlled within ±0.1MPa; when the steam supply temperature varies from 450℃ to 550℃, the prediction deviation does not exceed ±2℃.

[0090] The online data update mechanism established in step S2.6 solves the problem of model adaptability. Traditional offline training models are difficult to adapt to changes in system dynamic characteristics caused by factors such as equipment aging and changes in operating conditions. The present invention establishes a data foundation for dynamic updates by storing prediction results and actual parameters in real time. It has been verified that when the system has been running for more than 1,000 hours, the prediction accuracy of the model with an online update mechanism is about 15% higher than that of the fixed model. Especially under conditions of rapid load changes (load change rate exceeds 20% / hour), the improvement in prediction accuracy is more significant, reaching about 25%.

[0091] Through the above technical means, the present invention effectively solves two key problems in the practical application of the steam supply prediction model: first, it ensures the consistency of the prediction input and improves the generalization ability of the model; second, it realizes the dynamic optimization of the prediction model and enhances the adaptability of the system. These improvements have important engineering value for improving the intelligent level of the steam supply system of thermal power plants.

[0092] S3: The target steam supply parameters control the opening of the pipe network valves through the execution unit, so that the actual steam supply parameters of the thermal power plant steam supply network match the target steam supply parameters.

[0093] S3.1: Compare the target steam supply parameters with the real-time steam supply parameters, and calculate the steam supply pressure deviation, steam supply temperature deviation and steam supply flow deviation;

[0094] S3.2: Based on the deviation value, the fuzzy PID controller is used to calculate the target opening of each pipe network valve:

[0095] Based on the deviation value, a fuzzy rule set is established to fuzzify the steam supply pressure deviation, steam supply temperature deviation and steam supply flow deviation into five levels: {negative large, negative small, zero, positive small, positive large};

[0096] Design a fuzzy rule base to determine the adjustment coefficients of PID parameters according to the deviation level combination, where the adjustment coefficient range of Kp is [0.5, 1.5], the adjustment coefficient range of Ki is [0.3, 1.2], and the adjustment coefficient range of Kd is [0.4, 1.3];

[0097] If the steam supply pressure deviation is at the "negative large" level and the steam supply flow deviation is at the "positive large" level, the Kp adjustment coefficient is set to 1.5 and the Ki adjustment coefficient is set to 0.3;

[0098] The centroid method is used to defuzzify the fuzzy reasoning results and obtain the real-time parameter values ​​of the PID controller;

[0099] Calculate the incremental PID controller output according to the real-time parameter value;

[0100] The output is converted into the target opening value of each pipe network valve. If the change of the target opening value exceeds 2%, it is gradually adjusted in steps of 2%.

[0101] S33: If the calculated target opening change rate is greater than a preset opening change rate threshold, the target opening is adjusted step by step according to the preset opening change rate threshold;

[0102] S3.4: The execution unit receives the target opening and controls each pipe network valve to adjust the opening;

[0103] S3.5: Collect the adjusted actual steam supply parameters and calculate the matching degree with the target steam supply parameters;

[0104] S3.6: If the matching degree is less than the preset matching threshold and still fails to reach the preset matching threshold after three consecutive adjustments, an exception processing signal is triggered.

[0105] Preferably, the S3 step of the present invention mainly solves the technical difficulties of executing control in the steam supply system of a thermal power plant. In the traditional steam supply control system, due to the use of a single PID control strategy and fixed control parameters, it is difficult to cope with complex and changeable steam supply conditions. Especially in the face of large-scale load fluctuations, multi-valve collaborative control and other scenarios, there are problems such as slow adjustment process, difficult optimization of setting parameters, and delayed system response. The present invention realizes dynamic optimization of control parameters by introducing fuzzy rules to adaptively adjust PID parameters in S32, so that the controller can automatically adjust the control strategy according to different working conditions. Experimental data show that under the condition of severe load fluctuations (load change rate exceeds 20% / hour), compared with the traditional fixed parameter PID, the adjustment time is shortened by about 35% and the overshoot is reduced by about 40%. At the same time, the opening change rate limiting mechanism designed in S33 of the present invention effectively reduces valve wear while ensuring the control response speed by reasonably controlling the valve action speed, and reduces the number of valve actions by about 30%. In addition, the closed-loop feedback and exception handling mechanism constructed by S35 and S36 reduces the pressure fluctuation amplitude of the system by about 25% by real-time monitoring of the control effect and timely handling of abnormal situations. It is worth noting that these improvements are not the result of a simple superposition of various technical means, but are achieved through a carefully designed collaborative control strategy. For example, the combination of fuzzy PID control and opening rate limit not only ensures the rapidity of control, but also avoids frequent valve movements. This balance is of great value in practical engineering applications. Practice has proved that this control scheme not only improves the control accuracy and response speed of the steam supply system, but also extends the service life of key equipment, making a significant contribution to improving the overall operating efficiency of the steam supply system of thermal power plants.

[0106] In summary, the intelligent steam supply method proposed in the present invention realizes the intelligent control of the steam supply system through the synergistic effect of three key steps. In step S1, a three-layer LSTM network structure is used to process the time series data of the steam supply system. The long-term dependency of the system is effectively captured through the memory unit and the gating mechanism. Compared with the traditional feedforward neural network, the prediction error is reduced by about 15%; at the same time, the introduction of dynamic learning rate adjustment and early stop mechanism shortens the model training time by about 30%. In step S2, the generalization and adaptability problems of the prediction model are solved by unifying the feature vector format and the online data update mechanism, especially under the condition of rapid load change (load change rate exceeds 20% / hour), the prediction accuracy is improved by about 25%; the design of the double-layer safety protection mechanism effectively reduces the number of safety alarms by about 40%, and the fluctuation range of system parameters is reduced by about 25%. In step S3, the fuzzy PID control strategy dynamically optimizes the control parameters, shortens the adjustment time of the system by about 35% and reduces the overshoot by about 40% when the load fluctuates greatly; the opening change rate limiting mechanism effectively reduces valve wear and reduces the number of valve actions by about 30%. The realization of these improved effects depends on the organic combination of various technical means, rather than simple superposition. For example, the synergy between the high-precision output of the prediction model and the fuzzy PID control not only ensures the control accuracy, but also improves the dynamic response capability of the system. Practice has shown that the present invention not only improves the overall control performance of the steam supply system, but also extends the service life of key equipment, providing a feasible technical solution for the intelligent upgrade of the steam supply system of thermal power plants.

[0107] Embodiment 2 is an embodiment of the present invention, which provides an intelligent steam supply system for a thermal power plant, including: a training module, which is used to establish a steam supply training set according to historical steam supply parameters, and input the steam supply training set into a preset machine learning model for training to obtain a steam supply prediction model;

[0108] Output module, the steam supply prediction model receives the real-time steam supply parameters of the thermal power plant and outputs the target steam supply parameters based on the real-time steam supply parameters;

[0109] The target steam supply parameters control the opening of the pipe network valves through the execution unit, so that the actual steam supply parameters of the steam supply pipe network of the thermal power plant match the target steam supply parameters.

[0110] Example 3, reference Figure 2, is an embodiment of the present invention, which is different from the previous embodiment in that: if the function is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), disk or optical disk and other media that can store program codes.

[0111] The logic and / or steps represented in the flowchart or otherwise described herein, for example, can be considered as an ordered list of executable instructions for implementing logical functions, and can be embodied in any computer-readable medium for use by an instruction execution system, device or apparatus (such as a computer-based system, a system including a processor, or other system that can fetch instructions from an instruction execution system, device or apparatus and execute instructions), or in conjunction with such instruction execution systems, devices or apparatuses. For the purposes of this specification, "computer-readable medium" can be any device that can contain, store, communicate, propagate or transmit a program for use by an instruction execution system, device or apparatus, or in conjunction with such instruction execution systems, devices or apparatuses.

[0112] More specific examples of computer-readable media (a non-exhaustive list) include the following: an electrical connection with one or more wires (electronic device), a portable computer disk case (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable and programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disk read-only memory (CDROM). In addition, the computer-readable medium may even be a paper or other suitable medium on which the program is printed, since the program may be obtained electronically, for example, by optically scanning the paper or other medium, followed by editing, deciphering or, if necessary, processing in another suitable manner, and then stored in a computer memory.

[0113] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0114] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. An intelligent steam supply method for a thermal power plant, characterized in that: include: Establish a steam supply training set according to historical steam supply parameters, and input the steam supply training set into a preset machine learning model for training to obtain a steam supply prediction model; The steam supply prediction model receives the real-time steam supply parameters of the thermal power plant and outputs the target steam supply parameters based on the real-time steam supply parameters; The target steam supply parameter controls the opening of the pipe network valve via the execution unit, so that the actual steam supply parameter of the steam supply pipe network of the thermal power plant matches the target steam supply parameter.

2. The intelligent steam supply method for a thermal power plant according to claim 1, characterized in that: A steam supply training set is established according to historical steam supply parameters, and the steam supply training set is input into a preset machine learning model for training to obtain a steam supply prediction model, including the following steps: Collect the following historical steam supply parameters: steam supply pressure, steam supply temperature, pipe network valve opening, steam supply flow rate and heat load demand of the steam supply network; Dividing the historical steam supply parameters into time series according to the collection time; The historical steam supply parameters corresponding to each time point are constructed as a feature vector, and the steam supply pressure, steam supply temperature and steam supply flow rate at the next time point are constructed as a label vector; If the steam supply pressure in the feature vector is less than a preset pressure threshold or the steam supply temperature is less than a preset temperature threshold, the feature vector and the corresponding label vector are marked as abnormal samples; Combining the feature vectors and label vectors that are not marked as abnormal samples to form a steam supply training set; The steam supply training set is input into a preset long short-term memory network LSTM model for training to obtain a steam supply prediction model.

3. The intelligent steam supply method for a thermal power plant according to claim 2, characterized in that: The steam supply training set is input into a preset long short-term memory network LSTM model for training to obtain a steam supply prediction model, including the following steps: Dividing the steam supply training set into a training data set, a validation data set and a test data set in a ratio of 8:1:1; Construct a neural network structure with three layers of LSTM units, where the number of neurons in each layer of LSTM units is 128, 64, and 32 respectively; If the number of training rounds is less than 1000 and the validation set loss value has not decreased for 5 consecutive rounds, the learning rate is reduced to 0.8 times the original value; If the validation set loss value does not decrease for 10 consecutive rounds, the training process is stopped and the current model parameters are saved; Use the test data set to test the trained model and output the prediction accuracy.

4. The intelligent steam supply method for a thermal power plant according to claim 3, characterized in that: The steam supply prediction model receives the real-time steam supply parameters of the thermal power plant and outputs the target steam supply parameters based on the real-time steam supply parameters, including the following steps: Real-time collection of real-time steam supply parameters in the steam supply system of thermal power plants, including: real-time steam supply pressure, real-time steam supply temperature, real-time pipe network valve opening, real-time steam supply flow and real-time heat load demand; The real-time steam supply parameters are constructed into a feature vector in the same format as the step "constructing the historical steam supply parameters corresponding to each time sequence point into a feature vector, and constructing the steam supply pressure, steam supply temperature and steam supply flow rate at the next time sequence point into a label vector"; Inputting the feature vector into the trained steam supply prediction model; The steam supply prediction model outputs the predicted steam supply pressure, predicted steam supply temperature and predicted steam supply flow rate at the next time point, and these three parameters are used as target steam supply parameters; If the predicted steam supply pressure exceeds the safety pressure range or the predicted steam supply temperature exceeds the safety temperature range, the target steam supply parameter is set to a boundary value of the corresponding parameter range; The target steam supply parameters, real-time steam supply parameters and corresponding time information are stored in a historical database for subsequent online updating of the model.

5. The intelligent steam supply method for a thermal power plant according to claim 4, characterized in that: Setting the target steam supply parameter to a boundary value of a corresponding parameter range includes the following steps: Setting the target steam supply parameter to a boundary value of a corresponding parameter range includes: The safety range of steam supply pressure is set to [Pmin, Pmax], where Pmin is the minimum safety pressure value and Pmax is the maximum safety pressure value; the safety range of steam supply temperature is set to [Tmin, Tmax], where Tmin is the minimum safety temperature value and Tmax is the maximum safety temperature value; If the predicted steam supply pressure is less than Pmin, the target steam supply pressure is set to Pmin+ΔP, where ΔP is the pressure safety margin value; If the predicted steam supply pressure is greater than Pmax, the target steam supply pressure is set to Pmax-ΔP; If the predicted steam supply temperature is less than Tmin, the target steam supply temperature is set to Tmin+ΔT, where ΔT is the temperature safety margin value; If the predicted steam supply temperature is greater than Tmax, the target steam supply temperature is set to Tmax-ΔT; Calculate the deviation between the target steam supply parameters and the predicted steam supply parameters. If the deviation exceeds the preset threshold, an alarm signal is triggered and the alarm information is recorded.

6. The intelligent steam supply method for a thermal power plant according to claim 5, characterized in that: The target steam supply parameter controls the opening of the pipe network valve via the execution unit, so that the actual steam supply parameter of the steam supply pipe network of the thermal power plant matches the target steam supply parameter, including the following steps: Comparing the target steam supply parameter with the real-time steam supply parameter, and calculating the steam supply pressure deviation value, the steam supply temperature deviation value and the steam supply flow deviation value; Based on the deviation value, a fuzzy PID controller is used to calculate the target opening of each pipe network valve; If the calculated target opening change rate is greater than the preset opening change rate threshold, the target opening is adjusted step by step according to the preset opening change rate threshold; The execution unit receives the target opening and controls each pipe network valve to adjust the opening; Collect the actual steam supply parameters after adjustment and calculate the matching degree with the target steam supply parameters; If the matching degree is less than the preset matching threshold and still fails to reach the preset matching threshold after three consecutive adjustments, an abnormal processing signal is triggered.

7. The intelligent steam supply method for a thermal power plant according to claim 6, characterized in that: Based on the deviation value, a fuzzy PID controller is used to calculate the target opening of each pipe network valve, including the following steps: A fuzzy rule set is established based on the deviation value, and the steam supply pressure deviation, steam supply temperature deviation and steam supply flow deviation are fuzzified into five levels of {negative large, negative small, zero, positive small, positive large}; Design a fuzzy rule base to determine the adjustment coefficients of PID parameters according to the deviation level combination, where the adjustment coefficient range of Kp is [0.5, 1.5], the adjustment coefficient range of Ki is [0.3, 1.2], and the adjustment coefficient range of Kd is [0.4, 1.3]; If the steam supply pressure deviation is at the "negative large" level and the steam supply flow deviation is at the "positive large" level, the Kp adjustment coefficient is set to 1.5 and the Ki adjustment coefficient is set to 0.3; The centroid method is used to defuzzify the fuzzy reasoning results and obtain the real-time parameter values ​​of the PID controller; Calculate the incremental PID controller output according to the real-time parameter value; The output is converted into a target opening value of each pipe network valve. If the change of the target opening value exceeds 2%, it is gradually adjusted in steps of 2%.

8. An intelligent steam supply system for a thermal power plant, based on the intelligent steam supply method for a thermal power plant according to any one of claims 1 to 7, characterized in that: include, A training module, used to establish a steam supply training set according to historical steam supply parameters, and input the steam supply training set into a preset machine learning model for training to obtain a steam supply prediction model; An output module, wherein the steam supply prediction model receives real-time steam supply parameters of the thermal power plant and outputs target steam supply parameters based on the real-time steam supply parameters; The target steam supply parameter controls the opening of the pipe network valve via the execution unit, so that the actual steam supply parameter of the steam supply pipe network of the thermal power plant matches the target steam supply parameter.

9. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the intelligent steam supply method for a thermal power plant 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 intelligent steam supply method for a thermal power plant according to any one of claims 1 to 7 are implemented.

Images