Liquid slagging boiler waste heat utilization control method, system, equipment and medium

Abstract

The invention provides a liquid slagging boiler waste heat utilization control method, system, equipment and medium, and belongs to the technical field of liquid slagging boilers, and the method comprises the following steps: collecting a plurality of real-time operation parameters of a liquid slagging boiler; inputting the multiple pieces of real-time operation data into a pre-trained artificial neural network model, dynamically optimizing control parameters in the waste heat recovery process by using an optimization algorithm, searching an optimal control parameter combination in a solution space of the artificial neural network model, and generating a liquid slag discharge boiler control instruction; and the control instruction is used for adjusting operation parameters of all the waste heat recovery devices, and the waste heat utilization process of the liquid slagging boiler is dynamically controlled. The waste heat recovery efficiency of the liquid deslagging boiler can be improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of liquid slagging boilers, and in particular relates to a liquid slagging boiler waste heat utilization control method, system, equipment and medium. Background Art

[0002] Liquid slag boilers are widely used in industrial production, and the large amount of waste heat generated during their operation has enormous recovery and utilization value. However, traditional waste heat utilization technologies have many limitations. Firstly, the physical characteristics of liquid slag boilers, such as heat transfer and combustion processes, vary significantly under different operating conditions (such as startup, stable operation, load fluctuations, and different fuel types).

[0003] Traditional fixed-parameter control strategies are unable to adjust control parameters in real-time to changing operating conditions, thus impacting boiler efficiency, stability, and safety. This can easily lead to issues such as incomplete combustion, coking, overheating, and overpressure. Furthermore, in practical applications, multiple conflicting objectives must be considered, such as improving thermal efficiency, reducing pollutant emissions, minimizing equipment wear and maintenance costs, and ensuring stable steam quality. Traditional methods struggle to strike a reasonable balance between these multiple objectives, resulting in suboptimal overall operating efficiency and impacting waste heat recovery efficiency. Summary of the Invention

[0004] In order to overcome the above-mentioned deficiencies in the prior art, the present invention provides a method for controlling waste heat utilization of a liquid slag discharge boiler, comprising the following steps:

[0005] Collect multiple real-time operating parameters of liquid slag boilers;

[0006] Input multiple real-time operating data into a pre-trained artificial neural network model, use an optimization algorithm to search for optimal operating parameters of the waste heat recovery equipment within the solution space of the artificial neural network model, and generate control instructions for the waste heat recovery equipment;

[0007] Use control instructions to dynamically control the waste heat utilization process.

[0008] Preferably, before inputting the plurality of real-time operation data into the pre-trained artificial neural network model, the method further includes training the artificial neural network model, including the following steps:

[0009] Acquire multiple historical operating parameters of the liquid slag discharge boiler, input the multiple historical operating parameters into an artificial neural network model, determine the number of input layer nodes of the artificial neural network model according to the number of the multiple historical operating parameters, and set the loss function, number of iterations, and learning rate of the artificial neural network model;

[0010] In each iteration, forward propagation calculation is performed, linear transformation and activation function calculation are performed on each layer of the artificial neural network model, and the predicted output value of the operating parameters of the waste heat recovery equipment is obtained;

[0011] Based on the predicted output value and the actual output value, the value of the loss function is calculated. The backpropagation algorithm is used to calculate the gradient of the loss function with respect to the weights and biases of each layer of the artificial neural network model, and the gradient is backpropagated to the input layer. Based on the calculated gradient, the weights and biases are updated using the gradient descent method.

[0012] Repeat the above steps until the set number of iterations is reached.

[0013] Preferably, before training the artificial neural network model, the method further includes initializing the weights and bias items of neurons in the artificial neural network model using normal distribution to obtain initial values.

[0014] Preferably, before inputting the multiple historical operating parameters into the artificial neural network model, the method further includes cleaning the multiple historical operating parameters to remove abnormal values ​​and erroneous data.

[0015] Preferably, the artificial neural network model is a multilayer perceptron neural network model, which includes an input layer, an intermediate layer and an output layer. The number of layers in the intermediate layer is 1-3 hidden layers, and the number of nodes in each layer is 5-50.

[0016] Preferably, the multiple historical operating parameters of the liquid slag discharge boiler include: boiler load, liquid slag flow rate, liquid slag temperature, flue gas flow rate, flue gas temperature, air flow rate, air temperature and ambient temperature.

[0017] The present invention also provides a liquid slag discharge boiler waste heat utilization control system, comprising:

[0018] Data acquisition module, used to collect multiple real-time operating parameters of the liquid slag discharge boiler;

[0019] A data processing module is used to input multiple real-time operating data into a pre-trained artificial neural network model, use an optimization algorithm to search for optimal operating parameters of the waste heat recovery equipment within the solution space of the artificial neural network model, and generate control instructions for the waste heat recovery equipment;

[0020] The execution module is used to dynamically control the waste heat utilization process using control instructions.

[0021] The present invention also provides a computer device, comprising a memory and a processor; the memory stores a computer program, and the processor is used to run the computer program in the memory to execute the liquid slag discharge boiler waste heat utilization control method.

[0022] The present invention also provides a computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program, and the computer program is suitable for being loaded by a processor to execute the liquid slag discharge boiler waste heat utilization control method.

[0023] The liquid slag discharge boiler waste heat utilization control method provided by the present invention has the following beneficial effects:

[0024] The present invention inputs multiple operating data into a trained artificial neural network model and dynamically optimizes the waste heat recovery process by using an optimization algorithm. It can comprehensively consider the interrelationship and synergy of various parameters in the waste heat utilization control system of the liquid slag discharge boiler, thereby searching for the optimal operating parameters in the solution space of the artificial neural network model and generating control instructions for the waste heat recovery equipment. This process can optimize the energy distribution and transfer between the various parts of the waste heat recovery equipment, thereby ensuring that the system always operates in the optimal state, improving the waste heat recovery efficiency, and effectively reducing energy waste. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] To more clearly illustrate the embodiments of the present invention and its design, the following briefly introduces the drawings required for this embodiment. The drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be derived from these drawings without inventive effort.

[0026] Figure 1 This is a flow chart of a method for controlling waste heat utilization of a liquid slag discharge boiler according to an embodiment of the present invention.

[0027] Figure 2 It is the structural diagram of the artificial neural network model;

[0028] Figure 3 This is the control system diagram for waste heat utilization of liquid slag discharge boiler. DETAILED DESCRIPTION

[0029] In order to enable those skilled in the art to better understand the technical solution of the present invention and to be able to implement it, the present invention is described in detail below with reference to the accompanying drawings and specific embodiments. The following embodiments are only used to more clearly illustrate the technical solution of the present invention and are not intended to limit the scope of protection of the present invention.

[0030] In the description of the present invention, it should be understood that the terms "center", "longitudinal", "lateral", "length", "width", "thickness", "up", "down", "front", "back", "left", "right", "vertical", "horizontal", "top", "bottom", "inside", "outside", "axial", "radial", "circumferential" and the like to indicate orientations or positional relationships based on the orientations or positional relationships shown in the accompanying drawings, and are only for the convenience of describing the technical solutions of the present invention and simplifying the description, rather than indicating or implying that the device or element referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore should not be understood as limiting the present invention.

[0031] In addition, the terms "first", "second", etc. are used for descriptive purposes only and are not to be understood as indicating or implying relative importance. In the description of the present invention, it should be noted that, unless otherwise expressly specified or limited, the terms "connected" and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection or an indirect connection through an intermediate medium. For ordinary technicians in this field, the specific meaning of the above terms in the present invention can be understood according to the specific circumstances. In the description of the present invention, unless otherwise specified, "plurality" means two or more, which will not be described in detail here.

[0032] Example

[0033] The present invention provides a method for controlling waste heat utilization of a liquid slag discharge boiler, specifically Figure 1 As shown, the following steps are included:

[0034] Step 1: Collect multiple real-time operating parameters of the liquid slag discharge boiler.

[0035] The data acquisition system of the present invention is achieved by multiple high-precision sensors, including temperature sensors, pressure sensors, flow sensors, and liquid level sensors. The temperature sensors are thermocouples or RTD sensors, located at key locations of the liquid slag discharge boiler, such as the furnace outlet, flue, liquid slag cooling device, air preheater inlet and outlet, and water supply pipes. They can accurately monitor the temperature changes of high-temperature flue gas, liquid slag, and various media. The pressure sensors are piezoresistive sensors with a range customized according to different pipeline pressure requirements. They are installed in locations such as water supply pipes and flues to ensure accurate monitoring of the pressure of media such as water and air. The flow sensors are electromagnetic flowmeters or ultrasonic flowmeters, installed in liquid slag discharge pipes, water supply pipes, and air pipes, to accurately obtain real-time flow information of the media in each pipe. The liquid level sensors are non-contact ultrasonic liquid level sensors or contact static pressure liquid level sensors, effectively monitoring the liquid level height of containers such as water tanks. The data collected by the sensors is transmitted to the data processing module via wired (such as industrial Ethernet, RS485, etc.) or wireless (such as Wi-Fi, ZigBee, etc.) communication methods.

[0036] Step 2: Input multiple real-time operating data into the pre-trained artificial neural network model, use the optimization algorithm to search for the optimal operating parameters of the waste heat recovery equipment within the solution space of the artificial neural network model, and generate control instructions for the waste heat recovery equipment. The specific steps include:

[0037] (1) Train the artificial neural network model.

[0038] Multiple historical operating parameters are obtained and used as input parameters of the artificial neural network model. The historical operating parameters include relevant parameters such as boiler load, liquid slag and flue gas flow and temperature. The output parameters are the operating parameters of the waste heat recovery equipment (such as the inlet and outlet temperatures and flow rates of the equipment). By continuously adjusting the weights and biases of the neural network, the predicted operating parameters of each device can maximize the waste heat recovery efficiency η.

[0039] First, the acquired historical operating parameters are cleaned to remove outliers and erroneous data. Outliers may be caused by sensor failure or measurement errors under extreme operating conditions. For example, if a temperature sensor suddenly measures a value significantly outside the normal range, it can be identified as an outlier and replaced with a reasonable estimate.

[0040] Afterwards, the cleaned data is normalized to map all input and output data to a smaller range, usually between [0, 1] or [-1, 1]. Normalization can unify data of different scales and units, helping to improve the training speed and stability of the neural network and prevent certain features from dominating the training process due to excessively large or small values. For a data sequence, normalization uses the following formula:

[0041]

[0042] Among them, x and x normalized are the original data and the normalized data in a data sequence, respectively, x min and x max are the minimum and maximum values ​​in the data series respectively.

[0043] The processed data is divided into a training set (usually 70% to 80% of the total data), a validation set (10% to 15%), and a test set (10% to 15%). The training set is used to train the parameters of the neural network, the validation set is used to evaluate the performance of the model during training to prevent overfitting, and the test set is used to ultimately evaluate the trained model's ability to generalize to unseen data.

[0044] The artificial neural network model of the present invention selects a multi-layer perceptron neural network architecture, specifically as follows Figure 2 As shown in Figure 1, it includes an input layer, an intermediate layer (hidden layer), and an output layer. In a waste heat utilization control system for a liquid slag discharge boiler, the number of input layer nodes may include parameters such as boiler load, liquid slag flow rate, liquid slag temperature, flue gas flow rate, flue gas temperature, air flow rate, air temperature, and ambient temperature. The number of input layer nodes is determined based on the number of input parameters and the complexity of the problem.

[0045] The number of layers and nodes in the middle layer needs to be determined through experimentation and optimization. First, try 1 to 3 hidden layers (implicit layers), with 5 to 50 nodes per layer. For example, set up 2 hidden layers, with 20 nodes in the first layer and 10 nodes in the second layer. The activation function of the hidden layer is the ReLU function, which is expressed as:

[0046] f(x)=max(0,x);

[0047] The number of nodes in the output layer is determined by the target parameters to be predicted or optimized. In a waste heat utilization system, these output parameters include the optimal operating parameters of each waste heat recovery device (such as the target air outlet temperature of the air preheater, the target liquid slag outlet temperature of the liquid slag cooler, the water flow adjustment value of the economizer, etc.), as well as the performance indicators of the entire system (such as waste heat recovery efficiency and total system energy output). The activation function of the output layer is selected based on the range and properties of the output parameters. If the output is a continuous numerical range, a linear function is an appropriate choice; if the output needs to be limited to a certain range (such as a probability value between [0, 1]), a sigmoid function is used.

[0048] Suppose the input of the input layer is (x1, x2, x3), the output of the output layer is (y1, y2, y3), the output of the hidden layer is (h1, h2, h3, h4), and the weight matrix is:

[0049]

[0050]

[0051] Then, the calculation formula for the 4 neurons in the hidden layer is:

[0052] h1=f(w 11 x1+w 21 x2+w 31 x3)

[0053] h2=f(w 12 x1+w 22 x2+w 32 x3)

[0054] h3=f(w 13 x1+w 23 x2+w 33 x3)

[0055] h4=f(w 14 x1+w 24 x2+w 34 x3);

[0056] The calculation formula for the three neurons in the output layer is:

[0057] y1=f(w1h1+w2h2+w3h3+w4h4)

[0058] y2=f(w5h1+w6h2+w7h3+w8h4)

[0059] y3=f(w9h1+w 10 h2+w 11 h3+w 12 h4);

[0060] Among them, f represents the activation function, the parameters of the input layer are the working fluid temperature, pressure, flow rate, etc.; the parameters of the output layer are the regulating valve opening, driving pump speed, and fan motor inverter, etc.

[0061] Before training the neural network, the weights in the network need to be initialized. Weight initialization uses normal distribution to initialize the weights, that is, for each weight ω ij (the weight connecting the neurons in layer i and layer j), whose initial value can be obtained from the normal distribution N(0,σ 2 ), where σ is the standard deviation selected based on experience, usually between 0.01 and 0.1. j , the bias of the j-th layer neuron) can also be initialized to a small constant, such as 0 or a value close to 0.

[0062] Furthermore, a loss function is selected for model training. An appropriate loss function is selected based on the type of output parameters and the optimization goal. If the output is a continuous numerical prediction (such as temperature, flow, etc.), the commonly used loss function is the mean square error (MSE) function, which is expressed as:

[0063]

[0064] Among them, m is the number of training samples, y i is the true output value of the i-th sample, is the output value of the i-th sample predicted by the neural network.

[0065] If the output is a classification problem (for example, determining whether the waste heat utilization system is in an efficient operating state, whether there is a fault, etc.), a cross entropy loss function can be used.

[0066] Determine the number of training iterations between 100 and 1000. The specific value needs to be adjusted based on the complexity of the problem and the size of the data. Also, set the learning rate α to control the step size of each weight update. A learning rate that is too high may prevent the model from converging to the optimal solution, while a learning rate that is too low will make the training process very slow. Generally, try a value between 0.01 and 0.1 and then adjust it based on the convergence during training. For example, if the loss function of the model decreases too slowly during training, increase the learning rate appropriately. If the loss function oscillates or does not converge during training, decrease the learning rate.

[0067] Next, the neural network is trained using the training set data. In each iteration, the following steps are performed:

[0068] Input a batch of data from the training set into the neural network, perform forward propagation calculations, and obtain the predicted output. For example, for a batch containing b samples, input it into the feature matrix X b Input into the neural network, after linear transformation and activation function calculation of each layer, the output matrix is ​​obtained

[0069] According to the predicted output and the actual output Y b , calculate the value of the loss function. For example, for the mean square error loss function, calculate:

[0070]

[0071] The backpropagation algorithm is used to calculate the gradient of the loss function with respect to the weights and biases of each layer. Starting from the output layer, the gradient is calculated step by step according to the chain rule and the gradient is backpropagated to the input layer. For example, for the weight ω of the output layer L,j , and its gradient calculation formula is:

[0072]

[0073] where y b,i,j and are the true output and the jth element of the predicted output of the i-th sample in the b-th batch, respectively, f'(z b,i,j ) is the derivative of the output layer activation function, h b,i is the i-th element of the output vector of the last hidden layer.

[0074] Based on the calculated gradient, the weights and biases are updated using the gradient descent method. The weight update formula is:

[0075]

[0076] Repeat the above steps until the set number of iterations is reached or other stopping conditions are met.

[0077] During the training process, the validation set data is used to evaluate the performance of the model to prevent overfitting. In addition to the loss function, other evaluation indicators can also be selected, such as the root mean square error (RMSE):

[0078]

[0079] and the coefficient of determination (R 2 ):

[0080]

[0081] These indicators can reflect the deviation and degree of fit between the model's predicted value and the true value from different perspectives. If the model is found to have an overfitting problem on the validation set, a regularization term is added to the loss function, such as L1 regularization (the sum of the absolute values ​​of the weights) or L2 regularization (the sum of the squares of the weights). For example, for L2 regularization, the new loss function can be expressed as:

[0082]

[0083] Where is the regularization parameter, and its appropriate value needs to be determined through experiments. Regularization can prevent the weights from being too large, thereby reducing overfitting.

[0084] Furthermore, adjust the network structure, try to increase or decrease the number of hidden layers and nodes, retrain and evaluate the model, and find the optimal network structure.

[0085] Multiple real-time operating data are input into the trained artificial neural network model. The number of input layer nodes of the artificial neural network model is determined according to the number of operating parameters. The waste heat recovery process is dynamically optimized using an optimization algorithm. The optimal control parameter combination is searched in the solution space of the artificial neural network model to generate control instructions for the waste heat recovery equipment.

[0086] First, remove interference information such as noise and outliers in the real-time operation data and perform normalization.

[0087] Secondly, real-time operating data is fed into a trained artificial neural network model, and genetic algorithms, particle swarm optimization algorithms, and other techniques are used to search for the optimal control parameter combination within the solution space of the artificial neural network algorithm model. For example, when optimizing liquid slag waste heat recovery, parameters such as the heat transfer medium flow rate and flow rate in the liquid slag-water heat exchanger or liquid slag-thermal oil heat exchanger are adjusted to maximize heat transfer. For flue gas waste heat recovery, the matching relationship between the economizer feed water flow rate and the flue gas flow rate, as well as the heat exchange efficiency between air and flue gas in the air preheater, are optimized to ensure efficient flue gas waste heat recovery under various operating conditions.

[0088] Step 3: Use control instructions to dynamically control the waste heat utilization process.

[0089] Control commands are sent to the execution module, where the regulating valves precisely control the flow of media such as water and air in the pipelines according to the commands. For example, when the air flow to the air preheater needs to be increased to improve combustion efficiency, the electric regulating valve will increase its opening accordingly, allowing more cold air to enter the air preheater for heat exchange with the high-temperature flue gas. The pump speed is adjusted by a frequency converter to change the circulation flow and pressure of the liquid medium (such as water or thermal oil). For example, in a liquid slag waste heat recovery system, when the heat transfer efficiency of the liquid slag-thermal oil heat exchanger needs to be improved, the centrifugal pump speed will be increased to increase the circulation speed of the thermal oil and enhance heat transfer. The fan speed is also controlled by the frequency converter to adjust the air flow. For example, during boiler combustion, the speed of the axial flow fan is adjusted according to changes in fuel quantity and combustion conditions to provide the appropriate amount of combustion air to ensure sufficient and stable combustion.

[0090] Specifically, the actuator receives control commands from the data processing module and adjusts the operating parameters of various devices to achieve dynamic control of the waste heat utilization process of the liquid slag discharge boiler. The actuator includes regulating valves, pumps, fans, and frequency converters. The regulating valves, installed in water supply and air pipes, control the flow of the medium. The pumps drive the circulation of liquid media (such as water and thermal oil) in the waste heat recovery system. The fans adjust the air flow. The frequency converters control the motor speed, thereby adjusting the output power of the pumps and fans.

[0091] Specifically, the actuator regulating valve uses an electric regulating valve or a pneumatic regulating valve to precisely regulate the flow of media such as water and air. Pumps can be selected from centrifugal pumps, plunger pumps, or gear pumps based on the properties of the conveying medium and the flow head requirements. The pump has a wide flow regulation range and can achieve stepless speed regulation of 0-100% flow through a frequency converter. The fan uses a centrifugal fan or an axial flow fan, selected based on the air volume and pressure requirements. The fan speed can be adjusted within the range of 0-50Hz through a frequency converter to achieve precise control of the air volume. The frequency converter uses a vector control type frequency converter with high control accuracy. It can achieve soft starting, speed regulation, and energy-saving operation of the motor. When used in conjunction with the pump and fan, it can accurately adjust the operating parameters of the equipment according to the instructions of the data processing module.

[0092] Specifically, the waste heat recovery equipment primarily consists of a liquid slag waste heat recovery unit (not part of the boiler) and a flue gas waste heat recovery unit (part of the boiler). The liquid slag waste heat recovery unit includes a liquid slag cooler, a liquid slag-water heat exchanger, and a liquid slag-thermal oil heat exchanger. It recovers waste heat from the liquid slag through indirect or direct heat exchange and transfers the recovered heat to subsequent utilization systems. The flue gas waste heat recovery unit, equipped with an economizer, air preheater, and flue gas-water heat exchanger, utilizes waste heat from the flue gas to heat feed water, air, or other media.

[0093] Furthermore, the liquid-slag cooler in the liquid-slag waste heat recovery unit utilizes water or air cooling, offering high cooling efficiency and reducing the liquid-slag temperature from approximately 1500°C to 600-800°C. The liquid-slag-to-water heat exchanger utilizes a shell-and-tube or plate-type structure, offering a high heat transfer coefficient and effectively recovering heat from the liquid-slag, heating the water to 80-150°C. The liquid-slag-to-thermal oil heat exchanger is constructed from high-temperature-resistant alloy materials, ensuring reliable operation in high-temperature environments. It heats the thermal oil to 200-300°C, providing a heat source for subsequent organic Rankine cycle power generation or other thermal utilization.

[0094] Furthermore, the economizer in the flue gas waste heat recovery unit utilizes a finned-tube structure with a large heat exchange area, raising the feedwater temperature from ambient to 100-200°C, improving boiler thermal efficiency. The air preheater utilizes a rotary or tubular structure for efficient heat exchange between flue gas and air, preheating the air to 200-300°C. The tubular air preheater offers a simple structure and easy maintenance, while also effectively raising air temperature and improving combustion performance.

[0095] Specifically, during the operation of the liquid slag discharge boiler, the real-time collected operating parameter data is preprocessed and input into the trained neural network model. The model predicts the optimal control parameter values ​​under the current operating conditions (such as the optimal setting values ​​of fuel flow and air flow) based on the input data.

[0096] The control parameter values ​​output by the model are compared with the current actual parameter values ​​to calculate the control deviation. The boiler's operating parameters are then adjusted in real time through actuators (such as control valves and frequency converters) to reduce the control deviation and optimize the boiler's operating state. The above process of data collection, model prediction, and parameter adjustment is repeated at regular intervals to achieve dynamic, real-time control of boiler parameters. This allows the system to adapt to changes in operating conditions and the influence of internal and external interference factors during boiler operation, ensuring that the boiler always operates efficiently, stably, and safely.

[0097] Specifically, in the neural network algorithm, the input parameters include boiler load, liquid slag and flue gas flow and temperature and other related parameters, and the output is the operating parameters of the waste heat recovery equipment (such as the inlet and outlet temperatures and flow rates of the equipment). By continuously adjusting the weights and biases of the neural network, the predicted operating parameters of each device can maximize the waste heat recovery efficiency η.

[0098] Waste heat recovery efficiency is a key indicator to measure the performance of the waste heat recovery system. The calculation formula is: where Q recovered is the total amount of waste heat recovered, Q available It is the total amount of waste heat that can be utilized in the system. In the waste heat utilization control system of liquid slag boiler, the available waste heat mainly comes from high-temperature liquid slag and hot flue gas.

[0099] For liquid slag, the heat Q slag The formula Q slag =m slag C p,slag (T slag,in -T slag,out ) calculation, where m slag is the mass flow rate of liquid slag, C p,slag is the specific heat capacity of liquid slag, T slag,in and T slag,out are the temperatures of the liquid slag entering and leaving the waste heat recovery device; for the flue gas, its heat Q smoke Similarly, Q smoke =m smoke C p,smoke (T smoke,in -T smoke,out ) calculation, m smoke is the mass flow rate of flue gas, C p,smoke is the specific heat capacity of the flue gas, T smoke,in and T smoke,out is the temperature of the flue gas entering and leaving the relevant waste heat recovery equipment (such as air preheater). The total amount of waste heat recovered is the sum of the heat recovered by each waste heat recovery equipment (such as air preheater, liquid slag cooler, economizer, etc.), that is, where Q i is the heat recovered by the i-th waste heat recovery device.

[0100] The present invention also provides a liquid slag boiler waste heat utilization control system, such as Figure 3 As shown, including:

[0101] Data acquisition module, used to collect multiple real-time operating parameters of the liquid slag discharge boiler;

[0102] A data processing module is used to input multiple real-time operating data into a pre-trained artificial neural network model, use an optimization algorithm to search for optimal operating parameters of the waste heat recovery equipment within the solution space of the artificial neural network model, and generate control instructions for the waste heat recovery equipment;

[0103] The execution module is used to dynamically control the waste heat utilization process using control instructions.

[0104] The data processing module has built-in data cleaning units, data normalization units, data storage units, and data analysis units. The data cleaning unit is used to remove noise, outliers, and other interfering information from the collected data; the data normalization unit uniformly processes data of different ranges and units for subsequent algorithm analysis; the data storage unit uses a large-capacity hard disk array to ensure the security and reliability of long-term data storage; the data analysis unit is the core part, integrating a variety of artificial neural network algorithm models for in-depth mining and analysis of processed data, extracting data features, and establishing mapping relationships between data, thereby predicting the operating status and performance of the boiler waste heat utilization system and generating control strategies based on preset optimization goals. The data cleaning unit uses algorithms based on statistical methods and data rules, such as the box plot method to remove outliers and the moving average method to smooth data noise, to ensure the accuracy and reliability of the data; the data normalization unit uses minimum-maximum normalization or Z-score normalization methods to map data of different physical dimensions to a unified numerical range for easy algorithm processing. The data storage unit uses a relational database (such as MySQL, Oracle, etc.) or a non-relational database (such as MongoDB, Redis, etc.) to store historical data and real-time data. The data storage period can be set according to demand to facilitate long-term data analysis and model training.

[0105] The execution module is used to use control instructions to adjust the operating parameters of each waste heat recovery device and dynamically control the waste heat utilization process.

[0106] The liquid slag discharge boiler waste heat utilization control system of the present invention can also be integrated with the plant's energy management system (EMS) or distributed control system (DCS). Through a network communication interface, the boiler waste heat utilization system's operating data, performance indicators, optimized control strategies, and other information can be uploaded to the upper system, enabling comprehensive management and optimized scheduling of energy across the entire plant.

[0107] The present invention also provides a computer device, including a memory and a processor; the memory stores a computer program, and the processor is used to run the computer program in the memory to execute the liquid slag discharge boiler waste heat utilization control method.

[0108] The present invention also provides a computer-readable storage medium, characterized in that the computer-readable storage medium stores a computer program, and the computer program is suitable for being loaded by a processor to execute the liquid slag discharge boiler waste heat utilization control method.

[0109] The above-described embodiments are only preferred specific implementation methods of the present invention, and the protection scope of the present invention is not limited thereto. Any simple changes or equivalent replacements of the technical solutions that can be obviously obtained by any technician familiar with the field within the technical scope disclosed in the present invention fall within the protection scope of the present invention.

Claims

1. A method for controlling waste heat utilization of a liquid slag discharge boiler, characterized in that: The steps include: Collect multiple real-time operating parameters of liquid slag boilers; Input multiple real-time operating data into a pre-trained artificial neural network model, use an optimization algorithm to search for optimal operating parameters of the waste heat recovery equipment within the solution space of the artificial neural network model, and generate control instructions for the waste heat recovery equipment; Use control instructions to dynamically control the waste heat utilization process.

2. The liquid slag discharge boiler waste heat utilization control method according to claim 1, characterized in that: Before inputting the plurality of real-time operation data into the pre-trained artificial neural network model, the artificial neural network model is trained, including the following steps: Acquire multiple historical operating parameters of the liquid slag discharge boiler, input the multiple historical operating parameters into an artificial neural network model, determine the number of input layer nodes of the artificial neural network model according to the number of the multiple historical operating parameters, and set the loss function, number of iterations, and learning rate of the artificial neural network model; In each iteration, forward propagation calculation is performed, linear transformation and activation function calculation are performed on each layer of the artificial neural network model, and the predicted output value of the operating parameters of the waste heat recovery equipment is obtained; Based on the predicted output value and the actual output value, the value of the loss function is calculated. The backpropagation algorithm is used to calculate the gradient of the loss function with respect to the weights and biases of each layer of the artificial neural network model, and the gradient is backpropagated to the input layer. Based on the calculated gradient, the weights and biases are updated using the gradient descent method. Repeat the above steps until the set number of iterations is reached.

3. The liquid slag discharge boiler waste heat utilization control method according to claim 2, characterized in that: Before training the artificial neural network model, the method also includes initializing the weights and bias items of neurons in the artificial neural network model using normal distribution to obtain initial values.

4. The liquid slag discharge boiler waste heat utilization control method according to claim 2, characterized in that: Before inputting the multiple historical operating parameters into the artificial neural network model, the multiple historical operating parameters are cleaned to remove abnormal values ​​and erroneous data.

5. The liquid slag discharge boiler waste heat utilization control method according to claim 1, characterized in that: The artificial neural network model is a multilayer perceptron neural network model, which includes an input layer, an intermediate layer and an output layer. The number of layers in the intermediate layer is 1-3 hidden layers, and the number of nodes in each layer is 5-50.

6. The liquid slag discharge boiler waste heat utilization control method according to claim 2, characterized in that: The multiple historical operating parameters of the liquid slag discharge boiler include: boiler load, liquid slag flow rate, liquid slag temperature, flue gas flow rate, flue gas temperature, air flow rate, air temperature and ambient temperature.

7. A liquid slag discharge boiler waste heat utilization control system, characterized in that: include: Data acquisition module, used to collect multiple real-time operating parameters of the liquid slag discharge boiler; A data processing module is used to input multiple real-time operating data into a pre-trained artificial neural network model, use an optimization algorithm to search for optimal operating parameters of the waste heat recovery equipment within the solution space of the artificial neural network model, and generate control instructions for the waste heat recovery equipment; The execution module is used to dynamically control the waste heat utilization process using control instructions.

8. A computer device, characterized in that: It comprises a memory and a processor; the memory stores a computer program, and the processor is used to run the computer program in the memory to execute the liquid slag discharge boiler waste heat utilization control method according to any one of claims 1 to 6.

9. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, and the computer program is suitable for being loaded by a processor to execute the liquid slag discharge boiler waste heat utilization control method according to any one of claims 1 to 6.

Images