A method for controlling superheat degree of a low-temperature waste heat power generation system

Abstract

The application discloses a superheat degree control method of a low-temperature waste heat power generation system, and belongs to the technical field of low-temperature waste heat power generation system control engineering, and the method comprises the following steps: temperature sensors and pressure sensors are arranged in the low-temperature waste heat power generation system; sensor data is collected according to a preset sampling interval, and the data is preprocessed; a superheat degree calculator is used to calculate the superheat degree at each moment; a preset neural network model is trained to obtain a superheat degree prediction model; based on the superheat degree prediction model, an adaptive dynamic programming model is trained; wherein the adaptive dynamic programming model is used to determine the working medium pump rotating speed value at the current moment according to the superheat degree at the current moment; the trained adaptive dynamic programming model is deployed to an embedded platform as a controller; the controller and a PLC are used to control the superheat degree, so that the superheat degree reaches a set constant value. The technical scheme of the application has the advantages of strong adaptability, fast control speed, online learning of the algorithm and easy deployment.

Description

Technical Field

[0001] This invention relates to the field of control engineering technology for low-temperature waste heat power generation systems, and in particular to a method for controlling the superheat of low-temperature waste heat power generation systems based on an adaptive dynamic programming algorithm. Background Technology

[0002] Waste heat, as a typical secondary energy source, is closely related to various production processes in industrial society. It is the unused heat remaining in the energy conversion process of primary energy due to insufficient combustion and heat exchange. It is generated in large quantities in industrial production processes and discharged into the environment, causing thermal pollution.

[0003] To address these issues, countries worldwide have unanimously focused on developing new renewable energy sources and energy-saving and emission-reduction technologies to alleviate the energy crisis and environmental pressures. Low-temperature waste heat power generation systems, with their excellent performance and high energy conversion efficiency, are widely used in industrial production processes and even automobiles. In ORC (Organic Regenerative Thermal Power Generation) systems, the evaporator is a device that converts liquid working fluid into gaseous working fluid. The working fluid fully absorbs heat in the evaporator. When the absorbed heat causes the temperature of the working fluid to be lower than the saturation temperature under the current pressure, the superheat is negative; when the absorbed heat causes the temperature of the working fluid to be higher than the saturation temperature under the current pressure, the superheat is positive. Superheat has a significant impact on the stability and power generation efficiency of the ORC power generation system; therefore, maintaining the superheat of the low-temperature waste heat power generation system within a reasonable range is crucial. However, current methods for controlling the superheat of low-temperature waste heat power generation systems within a reasonable range suffer from drawbacks such as low control precision, slow response time, large steady-state error, weak stability, and difficulty in algorithm deployment. Summary of the Invention

[0004] This invention provides a method for controlling the superheat of a low-temperature waste heat power generation system, in order to solve the technical problems of existing methods for controlling the superheat of a low-temperature waste heat power generation system to maintain it within a reasonable range, such as low control accuracy, slow response time, large steady-state error, weak stability, and difficulty in algorithm deployment.

[0005] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0006] A method for controlling superheat in a low-temperature waste heat power generation system includes:

[0007] S1, a first temperature sensor and a first pressure sensor are installed at the heat source inlet of the low-temperature waste heat power generation system to detect the temperature and pressure at the heat source inlet; a second temperature sensor and a second pressure sensor are installed at the evaporator outlet of the low-temperature waste heat power generation system to detect the temperature and pressure at the evaporator outlet; a third temperature sensor is installed at the condenser inlet of the low-temperature waste heat power generation system to detect the temperature at the condenser inlet.

[0008] S2, collect detection data from the first temperature sensor, the first pressure sensor, the second temperature sensor, the second pressure sensor, and the third temperature sensor according to the preset sampling interval;

[0009] S3, preprocess the collected detection data;

[0010] S4. Based on the pre-processed detection data from the second temperature sensor and the second pressure sensor, the superheat value at the sampling time corresponding to the detection data is calculated using a superheat calculator.

[0011] S5. Using the pre-processed detection data from each sensor, the working fluid pump speed value, and the calculated superheat, a preset neural network model is trained to obtain a superheat prediction model.

[0012] S6. Based on the preprocessed detection data of each sensor and the superheat prediction model, train an adaptive dynamic programming model; wherein, the adaptive dynamic programming model is used to determine the working fluid pump speed value at the current time according to the superheat at the current time.

[0013] S7 deploys the trained adaptive dynamic programming model to the embedded platform as a controller.

[0014] S8, the superheat is controlled by the controller and PLC in linkage to make the superheat reach the set constant value.

[0015] Furthermore, the preprocessing of the collected detection data includes:

[0016] Remove data indicating that the low-temperature waste heat power generation system failed to operate, remove outlier data, remove data where the temperature at the evaporator outlet is higher than the heat source temperature, handle missing values, and normalize all data to 0-1.

[0017] Furthermore, the overheating calculator is obtained as follows:

[0018] Based on the saturation temperature and pressure reference table of the working fluid pump, the temperature and pressure in the reference table are fitted to determine the fitting coefficient and obtain the corresponding fitting function, which serves as a superheat calculator.

[0019] Furthermore, the preset neural network model is a BP neural network;

[0020] The process involves using preprocessed sensor data, working fluid pump speed, and calculated superheat to train a pre-defined neural network model, resulting in a superheat prediction model, including:

[0021] The pre-processed detection data from the first temperature sensor, the first pressure sensor, and the third temperature sensor, along with the working fluid pump speed value and the superheat value calculated using the superheat calculator for the previous moment, are selected as the input data for the BP neural network. The superheat value for the current moment is used as the output data for the BP neural network to train the weight matrix of the BP neural network. The trained BP neural network is then used as the superheat prediction model.

[0022] Furthermore, superheat is selected as the controller state variable, and the working fluid pump speed is selected as the controller input; the adaptive dynamic programming model consists of an evaluation network and an execution network. For the execution network, the normalized superheat at the current moment is selected as the input, and the working fluid pump speed at the current moment is selected as the output; for the evaluation network, the normalized superheat at the next moment and the working fluid pump speed at the next moment are selected as the input, and the output is the expected value of the cost function.

[0023] Furthermore, the utility function U(k) of the adaptive dynamic programming model is:

[0024] U(k) = k * (actual superheat output of the system - superheat setpoint) 2

[0025] Where k is the amplification factor, and the value of k is to avoid the evaluation network falling into gradient vanishing or gradient exploding.

[0026] Furthermore, the training of the adaptive dynamic programming model based on the preprocessed detection data from each sensor and the overheat prediction model includes:

[0027] Collect detection data from the second pressure sensor and the second temperature sensor, and normalize the collected detection data.

[0028] The normalized detection data is input into a low-pass filter for filtering. Then, the superheat calculator is used to calculate the superheat at the current moment based on the filtered detection data. The calculated superheat at the current moment is input into the execution network of the adaptive dynamic programming model to obtain the working fluid pump speed at the current moment.

[0029] The superheat at the next moment is obtained using the superheat prediction model, and the obtained superheat at the next moment is input into the execution network to obtain the working fluid pump speed at the next moment.

[0030] Input the superheat and the working fluid pump speed at the next moment into the evaluation network;

[0031] Calculate the execution network update error and update the weights of the execution network using the backpropagation algorithm;

[0032] Calculate the evaluation network update error and update the weights of the evaluation network using the backpropagation algorithm.

[0033] Furthermore, the superheat is controlled by the controller and PLC in linkage to bring the superheat to a set constant value, including:

[0034] S81, the PLC uses the second pressure sensor and the second temperature sensor to obtain the temperature and pressure data at the evaporator outlet, and sends the temperature and pressure data at the evaporator outlet to the controller.

[0035] S82, the controller calculates the superheat at the current moment using a superheat calculator based on the temperature and pressure data at the evaporator outlet, and then calculates the working fluid pump speed using an adaptive dynamic programming model based on the calculated superheat at the current moment, and then sends the calculated working fluid pump speed to the PLC.

[0036] S83, the PLC controls the working fluid pump according to the working fluid pump speed sent by the controller;

[0037] S84, cycle from S81 to S83 until the superheat reaches the set constant value.

[0038] Furthermore, after using the controller and PLC to control the superheat to reach a set constant value, the superheat control method for the low-temperature waste heat power generation system further includes:

[0039] Based on the actual operation of the low-temperature waste heat power generation system, an adaptive dynamic programming model is trained online.

[0040] The beneficial effects of the technical solution provided by this invention include at least the following:

[0041] The superheat control method for low-temperature waste heat power generation systems provided by this invention overcomes the shortcomings of existing technologies, such as low control accuracy, slow response time, large steady-state error, weak stability, and difficult algorithm deployment. The training process of the evaluation network and execution network of the adaptive dynamic programming algorithm can be carried out online based on production data, enabling training without affecting normal production. The adaptive dynamic programming algorithm is a data-driven control algorithm that solves the problem of excessive dependence on system modeling in traditional control algorithms. Therefore, the system superheat control achieved based on this method has small error and fast response speed.

[0042] The superheat control method for low-temperature waste heat power generation system proposed in this invention can realize real-time control of the superheat of the low-temperature waste heat power generation system, realize the information acquisition and information processing functions of the sensor, and has high control accuracy, strong stability, easy deployment, and high control frequency, which can meet the control requirements of the superheat of the low-temperature waste heat power generation system. Attached Figure Description

[0043] To more clearly illustrate the technical solutions in the embodiments of the present invention, the accompanying drawings used 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 those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0044] Figure 1 This is a flowchart of the superheat control method for a low-temperature waste heat power generation system provided in an embodiment of the present invention;

[0045] Figure 2 This is an example diagram showing the placement of temperature and pressure sensors in the low-temperature waste heat power generation system provided in this embodiment of the invention.

[0046] Figure 3 This is a schematic diagram of the joint control of the embedded controller and PLC provided in an embodiment of the present invention;

[0047] Figure 4 This is a flowchart of the online training process for the adaptive dynamic programming algorithm provided in this embodiment of the invention. Detailed Implementation

[0048] To make the objectives, technical solutions, and advantages of the present invention clearer, the embodiments of the present invention will be described in further detail below with reference to the accompanying drawings.

[0049] First, it should be noted that in the embodiments of the present invention, the words "exemplarily," "for example," etc., are used to indicate that they are examples, illustrations, or descriptions. Any embodiment or design scheme described as "exemplary" in the present invention should not be construed as being more preferred or advantageous than other embodiments or design schemes. Specifically, the use of the term "exemplarily" is intended to present the concept in a specific manner. Furthermore, in the embodiments of the present invention, the meaning expressed by "and / or" can be both, or it can be either one or the other.

[0050] Furthermore, in the embodiments of the present invention, the terms "image" and "picture" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, their intended meanings are consistent. Similarly, the terms "of," "corresponding (relevant)," and "corresponding" may sometimes be used interchangeably. It should be noted that, without emphasizing the distinction between them, their intended meanings are consistent.

[0051] Furthermore, in embodiments of the present invention, sometimes a subscript (such as W1) may be mistakenly written as a non-subscript form (such as W1). Without emphasizing the difference, the meaning they express is the same.

[0052] To address the challenge of controlling the superheat at the evaporator outlet of a low-temperature waste heat power generation system and maintain the superheat at the evaporator outlet at a consistently set value, this embodiment provides a method for controlling the superheat in such a system. This method collects data from the system's temperature and pressure sensors and uses an Adaptive Dynamic Programming (ADP) algorithm to control the superheat at the evaporator outlet. An embedded platform (ARM or DSP platform) is used as the controller, serving as the computational platform for the ADP algorithm. Through coordinated control between the embedded controller and a PLC, the PLC receives information calculated by the embedded controller and controls the working fluid pump speed, thereby maintaining a constant superheat in the system. Specifically, the execution flow of this method is as follows: Figure 1 As shown, it includes the following steps:

[0053] S1, multiple temperature sensors and pressure sensors are installed in the low-temperature waste heat power generation system;

[0054] Specifically, such as Figure 2 As shown, this embodiment includes two sensors at the heat source inlet: a pressure sensor PT101 and a temperature sensor TE101, to detect the pressure and temperature at the heat source inlet. Two sensors at the system evaporator outlet are also included: a temperature sensor TE301 and a pressure sensor PT301, to detect the temperature and pressure at the evaporator outlet. A temperature sensor TE201 is installed at the condenser inlet to detect the temperature at the condenser inlet. Collecting the temperature and pressure information at the heat source inlet allows the neural network to determine how much heat is entering the system at any given moment. Collecting the temperature information of the condensate at the condenser inlet and maintaining a constant speed of the chilled water pump allows the neural network to determine how much heat is being lost. Collecting the temperature and pressure at the evaporator outlet allows for the calculation of the system's superheat at any given moment.

[0055] S2, collect the detection data of each sensor according to the preset sampling interval;

[0056] S3, preprocess the collected detection data;

[0057] Specifically, this embodiment achieves data cleaning through data preprocessing, including: data denoising: detecting and removing noise, incorrect labels, outliers, or inconsistent data from the data. Handling missing values ​​in the data: this can be achieved by deleting samples containing missing values, interpolation imputation, or other methods. Deleting data indicating system malfunction, deleting outlier data, deleting data where the evaporator outlet temperature is higher than the heat source temperature, performing 0-1 normalization on all data, and data misalignment merging. Data misalignment merging is used to shift the superheat at the current moment to the previous moment, so that the system superheat data can be used as input to the BP neural network in subsequent steps for training the BP neural network.

[0058] S4. Based on the pre-processed data from temperature sensor TE301 and pressure sensor PT301, the superheat corresponding to the sampling time of the corresponding detection data is calculated using a superheat calculator.

[0059] Specifically, in this embodiment, the method for calculating superheat in S4 above is as follows:

[0060] The superheat calculator for the training system employs a function fitting method. Based on a table of saturation temperature and pressure of the working fluid, the values ​​of a, b, and c in the formula f(x) = a*x^b + c are trained on a computer to obtain the fitting function. Using the computer-trained formula and data from temperature sensor TE301 and pressure sensor PT301, the superheat of the system at each moment is calculated.

[0061] S5. Using the pre-processed detection data from each sensor, the working fluid pump speed value, and the calculated superheat, a preset neural network model is trained to obtain a superheat prediction model.

[0062] Specifically, in this embodiment, the neural network in S5 is a BP neural network. When training this BP neural network, this embodiment merges the data from temperature sensors TE101, TE201, and TE301 with the previous superheat data SHPRE. The data from TE101, TE201, TE301, the working fluid pump speed, and the SHPRE data are selected as the input to the BP neural network, and the current superheat data SH is selected as the output to train the weight matrix of the BP neural network. In this embodiment, the input and output data of the BP neural network are normalized to 0-1. The hidden layer of the BP neural network uses the ReLU activation function, the output layer uses a linear function, and the number of neurons in the hidden layer is selected as 13. The backpropagation algorithm is used to adjust the weights.

[0063] The functional expression for ReLU is:

[0064] f(x) = max(x)

[0065] S6. Based on the preprocessed detection data of each sensor and the superheat prediction model, train an adaptive dynamic programming model; wherein, the adaptive dynamic programming model is used to determine the working fluid pump speed value at the current time according to the superheat at the current time.

[0066] In this embodiment, the system superheat is selected as the controller state variable, and the working fluid pump speed is the controller input. The adaptive dynamic programming controller consists of an evaluation network and an execution network. For the execution network, the normalized current superheat is selected as its input, and the working fluid pump speed at the next moment is selected as its output. For the evaluation network, the normalized superheat information at the next moment and the working fluid pump speed at the next moment are selected as inputs, and the output is the expected value of the cost function, which is the accumulation of control error.

[0067] The training error of the execution network is:

[0068] ea=γj(k)-(j(k-1)+U(k)

[0069] In the formula, the parameter γ is the discount factor. When the discount factor is equal to 0, only the utility function at the current time is used. When the discount factor is equal to 1, all utility functions at future time points are considered equally important. The smaller the discount factor, the easier it is for the experiment to succeed; the larger the discount factor, the better the control effect. Therefore, the discount factor should not be too large. U(k) in the formula is a very important function called the utility function. The greater the difference between the actual superheat of the system and the set superheat, the larger the value of the utility function. The selection of the utility function has a significant impact on the stability, speed, and accuracy of the adaptive dynamic programming algorithm.

[0070] In this embodiment, the selected utility function is:

[0071] U(k) = k * (actual superheat output of the system - superheat setpoint) 2

[0072] Here, k is the amplification factor, and the value of k is used to avoid the evaluation network falling into the gradient vanishing or gradient exploding traps. The value of k is selected based on the learning performance of the network.

[0073] Furthermore, the learning rate has a significant impact on the stability of neural networks. A larger learning rate leads to faster learning, but an excessively large learning rate can cause oscillations, while an excessively small learning rate results in slow learning and excessively long training times. Therefore, the learning rate is typically decreased from a large initial value to a smaller value over time to accelerate learning and avoid oscillations.

[0074] The sigmoid function is used to evaluate the hidden layers of the network, and a linear network is used for the output layer. Its expression is:

[0075] inputs c =[x1(k) … x n (k) u]

[0076]

[0077]

[0078] J(k) = p·W c (2)

[0079] The sigmoid function is used to evaluate the hidden layers of the network, and the sigmoid function is used for the output layer. The expression is as follows:

[0080] inputs a =[x1(k) … x n (k)]

[0081] h = inputs a ·W c (1)

[0082]

[0083] v = g·W c (1)

[0084]

[0085] Based on the above, it should be noted that for the training of the adaptive dynamic programming controller, that is, the training of the controller evaluation network and the execution network, in order to avoid gradient vanishing and gradient explosion during the training process, the data of the input layer must be within a reasonable range. This is because the hidden layer of the controller neural network in this embodiment uses the sigmoid function, and the gradient sensitive range of this function is -5 to 5. Therefore, the input data of the neural network must be between -5 and 5, so the data must be normalized.

[0086] S7 deploys the trained adaptive dynamic programming model to the embedded platform as a controller.

[0087] The embedded platform can be either an ARM platform or a DSP platform. Deploying the adaptive dynamic programming model to the embedded platform involves writing the adaptive dynamic programming algorithm in C or Python, enabling floating-point operations, and testing the code's functionality. Computing neural networks requires a large number of floating-point operations; therefore, an external embedded platform will provide better control performance. The algorithm, the trained execution network, and the weight matrices of the evaluation network are then written into the embedded controller (ARM platform or DSP platform). An ARM platform refers to a chip with an ARM architecture CPU; high-performance ARM architecture chips have high clock speeds and floating-point operation capabilities. A DSP platform refers to a signal processing chip, which is an integrated circuit specifically designed for processing digital signals. These chips are typically used for real-time digital signal processing.

[0088] S8, the superheat is controlled by the controller and PLC in linkage to make the superheat reach the set constant value.

[0089] Specifically, in this embodiment, S8 includes the following steps:

[0090] The S81 PLC uses pressure and temperature sensors to obtain temperature and pressure data at the evaporator outlet and sends the data to the embedded controller via cable.

[0091] The S82 embedded controller calculates the speed of the working fluid pump based on an adaptive dynamic programming algorithm, and then sends the working fluid pump speed information to the PLC.

[0092] S83, the PLC controls the working fluid pump based on the working fluid pump speed information sent by the embedded controller;

[0093] S84, cycle from S81 to S83, until the superheat reaches the set constant value.

[0094] Among them, such as Figure 3 As shown, the process of superheat control by the embedded controller and PLC can be summarized as follows: the embedded controller acts as the system's algorithm processing platform, while the PLC acts as the sensor data collection and execution platform. First, the temperature sensor TE301 and pressure sensor PT301 are connected to the PLC's analog input port. The PLC collects the temperature and pressure information at the evaporator outlet in real time. Then, the PLC analyzes the temperature and pressure at the evaporator outlet and sends them to the embedded controller via cable. The communication protocol is RS485 (the PLC has an integrated RS485 interface; the embedded controller requires an external MAX485 chip). After receiving the temperature and pressure signals, the embedded controller must first use a low-pass filter to filter out high-frequency interference signals. Then, it uses a trained superheat calculator to calculate the current system superheat. Next, the embedded controller uses an adaptive dynamic programming algorithm to calculate the appropriate speed to be applied to the working fluid pump. After calculating the working fluid pump speed, the embedded controller transmits the pump speed information back to the PLC via cable. Upon receiving the pump speed information, the PLC then issues instructions to the working fluid pump. This process is repeated until the system superheat is constant, thus achieving superheat control.

[0095] Through the above scheme, this embodiment combines an embedded controller and a PLC to control overheating. The embedded controller is responsible for calculating the control algorithm, while the PLC is responsible for collecting sensor information and issuing commands to the working fluid pump. This structure allows the control system to be configured with only a low-performance PLC. Compared to the traditional method of using a high-performance PLC to run complex algorithms, this embodiment reduces construction costs and achieves stronger computing performance. More importantly, this scheme allows an embedded controller to be directly connected in parallel with the original system without requiring extensive modifications to the original system's PLC code.

[0096] Furthermore, the method in this embodiment also includes: training the weights of the evaluation network and the execution network of the adaptive dynamic programming algorithm online according to the actual running conditions.

[0097] The embedded controller can use an online-trained adaptive dynamic programming algorithm, or it can retrain neural network weights that are more suitable for the actual system based on the online-trained evaluation network and execution network weights to achieve better control performance.

[0098] Specifically, the online training process of the adaptive dynamic programming algorithm is as follows: Figure 4 As shown, the online training steps include: (1) collecting data from pressure sensor PT301 and temperature sensor TE301, and normalizing the data; (2) passing the data through a low-pass filter for filtering, and then passing it through the execution network of the controller to obtain the working fluid pump speed at this moment, and sending the speed information to the PLC; (3) obtaining the superheat information at the next moment, passing it through the execution network to obtain the working fluid pump speed at the next moment, and sending the speed information to the PLC; (4) calculating the execution network update error, updating the weights of the execution network using the backpropagation algorithm; calculating the evaluation network update error, and updating the weights of the evaluation network using the backpropagation algorithm.

[0099] In summary, this embodiment provides a method for controlling superheat in a low-temperature waste heat power generation system, overcoming the shortcomings of existing technologies such as low control accuracy, slow response time, large steady-state error, weak stability, and difficult algorithm deployment. The training process of the evaluation network and execution network of the adaptive dynamic programming algorithm can be performed online based on production data, enabling training without affecting normal production. The adaptive dynamic programming algorithm is a data-driven control algorithm that solves the problem of excessive dependence on system modeling in traditional control algorithms. Therefore, the system superheat control achieved based on this method has small error and fast response speed.

[0100] Furthermore, it should be noted that the present invention can be provided as a method, apparatus, or computer program product. Therefore, embodiments of the present invention can take the form of a completely or partially hardware embodiment, a completely or partially software embodiment, or an embodiment combining software and hardware aspects. Moreover, when implemented in software, embodiments of the present invention can take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program code. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer program are loaded or executed on a computer, all or part of the processes or functions described in the embodiments of the present invention are generated. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium or transmitted from one computer-readable storage medium to another. For example, the computer instructions can be transmitted from one website, computer, server, or data center to another website, computer, server, or data center via wired (e.g., infrared, wireless, microwave, etc.) means. The computer-readable storage medium can be any usable medium accessible to a computer or a data storage device such as a server or data center containing one or more sets of usable media. The available medium can be a magnetic medium (e.g., floppy disk, hard disk, magnetic tape), an optical medium (e.g., DVD), or a semiconductor medium. A semiconductor medium can be a solid-state drive (SSD).

[0101] Embodiments of the present invention are described with reference to flowchart illustrations and / or block diagrams of methods, terminal devices (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, embedded processor, or other programmable data processing terminal device to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing terminal device, generate instructions for implementing the flowchart illustrations. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0102] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing terminal device to operate in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1The functions specified in one or more boxes. These computer program instructions may also be loaded onto a computer or other programmable data processing terminal equipment to cause a series of operational steps to be performed on the computer or other programmable terminal equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable terminal equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0103] It should also be noted that, in this document, relational terms such as "first" and "second" are used only to distinguish one entity or operation from another, and do not necessarily require or imply any such actual relationship or order between these entities or operations. The terms "comprising," "including," or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or terminal device that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such a process, method, article, or terminal device. Without further limitations, an element defined by the phrase "comprising one..." does not exclude the presence of other identical elements in the process, method, article, or terminal device that includes said element. Furthermore, the term "and / or" is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A alone, A and B simultaneously, and B alone, where A and B can be singular or plural. Additionally, the character " / " in this text generally indicates an "or" relationship between the preceding and following objects, but it can also indicate an "AND / OR" relationship. Please refer to the context for specific interpretations. "At least one" refers to one or more items, while "more than" refers to two or more items. "At least one of the following" or similar expressions refer to any combination of these items, including any combination of single or multiple items. For example, at least one of a, b, or c can be represented as: a, b, c, ab, ac, bc, or abc, where a, b, and c can be single or multiple.

[0104] Furthermore, it is understood that in various embodiments of the present invention, the order of the above-mentioned process numbers does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present invention.

[0105] Those skilled in the art will recognize that the units and algorithm steps of the various examples described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are implemented in hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementations should not be considered beyond the scope of this invention.

[0106] In the several embodiments provided by this invention, it should be understood that the disclosed devices, apparatuses, and methods can be implemented in other ways. For example, the apparatus embodiments described above are merely illustrative. For instance, the division of functional modules / units is only a logical functional division, and in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another device, or some features may be ignored or not executed. Furthermore, the shown or discussed mutual couplings or direct couplings or communication connections may be through some interfaces; indirect couplings or communication connections between devices or units may be electrical, mechanical, or other forms. Units described as separate components may or may not be physically separate, and components shown as units may or may not be physical units, i.e., they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs. Additionally, the functional units in the various embodiments of this invention may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.

[0107] If the method is implemented as 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 a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods described in the various embodiments of the present invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0108] Finally, it should be noted that the above description represents a preferred embodiment of the present invention. It should be pointed out that although preferred embodiments have been described, those skilled in the art, once they understand the basic inventive concept of the present invention, can make various improvements and modifications without departing from the principles described herein. These improvements and modifications should also be considered within the scope of protection of the present invention. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the embodiments of the present invention.

Claims

1. A method for controlling superheat in a low-temperature waste heat power generation system, characterized in that, include: S1, A first temperature sensor and a first pressure sensor are installed at the heat source inlet of the low-temperature waste heat power generation system to detect the temperature and pressure at the heat source inlet. A second temperature sensor and a second pressure sensor are installed at the evaporator outlet of the low-temperature waste heat power generation system to detect the temperature and pressure at the evaporator outlet; a third temperature sensor is installed at the condenser inlet of the low-temperature waste heat power generation system to detect the temperature at the condenser inlet. S2, collect detection data from the first temperature sensor, the first pressure sensor, the second temperature sensor, the second pressure sensor, and the third temperature sensor according to the preset sampling interval; S3, preprocess the collected detection data; S4. Based on the pre-processed detection data from the second temperature sensor and the second pressure sensor, the superheat value at the sampling time corresponding to the detection data is calculated using a superheat calculator. S5. Using the pre-processed detection data from each sensor, the working fluid pump speed value, and the calculated superheat, a preset neural network model is trained to obtain a superheat prediction model. S6. Based on the preprocessed detection data of each sensor and the superheat prediction model, train an adaptive dynamic programming model; wherein, the adaptive dynamic programming model is used to determine the working fluid pump speed value at the current time according to the superheat at the current time. S7 deploys the trained adaptive dynamic programming model to the embedded platform as a controller. S8, the superheat is controlled by the controller and PLC in linkage to make the superheat reach the set constant value; The superheat is selected as the controller state variable, and the working fluid pump speed is the controller input. The adaptive dynamic programming model consists of an evaluation network and an execution network. For the execution network, the normalized superheat at the current moment is selected as the input, and the working fluid pump speed at the current moment is selected as the output. For the evaluation network, the normalized superheat at the next moment and the working fluid pump speed at the next moment are selected as the input, and the output is the expected value of the cost function. The utility function of the adaptive dynamic programming model... U ( k )for: ； Where k is the amplification factor, and the value of k is to avoid the evaluation network from falling into gradient vanishing or gradient exploding; The training of an adaptive dynamic programming model based on the preprocessed detection data from each sensor and the overheat prediction model includes: Collect detection data from the second pressure sensor and the second temperature sensor, and normalize the collected detection data. The normalized detection data is input into a low-pass filter for filtering. Then, the superheat calculator is used to calculate the superheat at the current moment based on the filtered detection data. The calculated superheat at the current moment is input into the execution network of the adaptive dynamic programming model to obtain the working fluid pump speed at the current moment. The superheat at the next moment is obtained using the superheat prediction model, and the obtained superheat at the next moment is input into the execution network to obtain the working fluid pump speed at the next moment. Input the superheat and the working fluid pump speed at the next moment into the evaluation network; Calculate the execution network update error and update the weights of the execution network using the backpropagation algorithm; Calculate the evaluation network update error and update the weights of the evaluation network using the backpropagation algorithm.

2. The superheat control method for a low-temperature waste heat power generation system as described in claim 1, characterized in that, The preprocessing of the collected detection data includes: Remove data indicating that the low-temperature waste heat power generation system failed to operate, remove outlier data, remove data where the temperature at the evaporator outlet is higher than the heat source temperature, handle missing values, and normalize all data to 0-1.

3. The superheat control method for a low-temperature waste heat power generation system as described in claim 1, characterized in that, The overheating calculator is obtained as follows: Based on the saturation temperature and pressure reference table of the working fluid pump, the temperature and pressure in the reference table are fitted to determine the fitting coefficient and obtain the corresponding fitting function, which serves as a superheat calculator.

4. The superheat control method for a low-temperature waste heat power generation system as described in claim 1, characterized in that, The preset neural network model is a BP neural network; The process involves using preprocessed sensor data, working fluid pump speed, and calculated superheat to train a pre-defined neural network model, resulting in a superheat prediction model, including: The pre-processed detection data from the first temperature sensor, the first pressure sensor, and the third temperature sensor, along with the working fluid pump speed value and the superheat value calculated using the superheat calculator for the previous moment, are selected as the input data for the BP neural network. The superheat value for the current moment is used as the output data for the BP neural network to train the weight matrix of the BP neural network. The trained BP neural network is then used as the superheat prediction model.

5. The superheat control method for a low-temperature waste heat power generation system as described in claim 1, characterized in that, The superheat is controlled by the controller and PLC in linkage to bring the superheat to a set constant value, including: S81, the PLC uses the second pressure sensor and the second temperature sensor to obtain the temperature and pressure data at the evaporator outlet, and sends the temperature and pressure data at the evaporator outlet to the controller. S82, the controller calculates the superheat at the current moment using a superheat calculator based on the temperature and pressure data at the evaporator outlet, and then calculates the working fluid pump speed using an adaptive dynamic programming model based on the calculated superheat at the current moment, and then sends the calculated working fluid pump speed to the PLC. S83, the PLC controls the working fluid pump according to the working fluid pump speed sent by the controller; S84, cycle from S81 to S83 until the superheat reaches the set constant value.

6. The method for controlling superheat in a low-temperature waste heat power generation system as described in any one of claims 1 to 5, characterized in that, After using the controller and PLC to control the superheat to a set constant value, the superheat control method of the low-temperature waste heat power generation system further includes: Based on the actual operation of the low-temperature waste heat power generation system, an adaptive dynamic programming model is trained online.

Images