Operation optimization method and system based on desulfurization system

Abstract

The invention provides an operation optimization method based on a desulfurization system and a system thereof, and relates to the technical field of desulfurization equipment operation control. The health coefficient of a pump is obtained by calculating the real-time stability and the performance loss degree of a liquid circulating pump and an oxygen pump, the health degree is judged, the blockage coefficient of a spraying system is calculated, and the operation optimization of the system is achieved. The equipment failure rate of the desulfurization system is calculated and judged by setting a blockage threshold value and the health degree of a pump, historical operation parameter data is screened to form a training data set according to the cost ratio of the desulfurization system and the equipment failure rate, and an operation optimization model of the desulfurization system is constructed through a long-short-term memory neural network. The method comprises the steps that an operation optimization model is established, the operation optimization model is trained, then real-time detection parameters are collected to serve as input of the model, an optimization result is output through the optimization model, and the operation optimization model is optimized through a sparrow search algorithm and according to a loss function serving as a fitness evaluation index.

Description

Technical Field

[0001] The present invention relates to the technical field of desulfurization equipment operation control, and in particular to a method and system for operation optimization based on a desulfurization system. Background Art

[0002] In the past few decades, coal has been one of the main pillars of industrialization and energy development in various countries, but it also faces serious environmental problems. The sulfur dioxide produced during the combustion of coal not only has a serious impact on the environment, but also poses a threat to human health. In the traditional method of flue gas desulfurization, it relies too much on manual control, which often causes problems such as substandard emissions and excessive costs. How to optimize and improve the desulfurization system has become an urgent problem to be solved.

[0003] In the prior art, publication number CN115355163A discloses a method for optimizing and managing a desulfurization system. A data-fused desulfurization efficiency model is established through the operation data of the desulfurization system. The model is used in the actual production process to optimize the desulfurization system in real time. Under the premise of ensuring the desulfurization efficiency, the optimal slurry circulation pump combination is output for the on-site operator to adjust the slurry circulation pump combination to reduce the total current of the slurry circulation pump and the total power consumption of the desulfurization system. The start and stop time intervals and the accumulated operating time of the slurry circulation pump are counted through the edge management device. At the same time, the health status of the slurry circulation pump is monitored to ensure the long-term continuous and normal operation of the slurry circulation pump, thereby solving the problem that the empirical adjustment method is relatively extensive and causes large desulfurization power consumption. However, the scheme only considers the energy consumption of the slurry circulation pump, ignores the use cost of the desulfurizer and the cost of system downtime caused by equipment failure, and lacks optimization of the overall parameters of the desulfurization system.

[0004] The above information disclosed in this Background section is only for enhancement of understanding of the background of the present disclosure and therefore it may contain information that does not constitute the prior art that is already known to one of ordinary skill in the art. Summary of the invention

[0005] The object of the present invention is to provide a method and system based on operation optimization of a desulfurization system to solve the problems raised in the above background technology.

[0006] To achieve the above object, the present invention provides the following technical solutions:

[0007] A method and system for optimizing the operation of a desulfurization system, the specific steps of which include:

[0008] Step 1: Obtain the historical data of current, vibration, temperature and noise of the desulfurization system pump to be optimized, including the slurry circulation pump and the oxygen pump, calculate the stability and performance loss degree of the pump, obtain the health factor of the pump, judge the health of the pump according to the health factor, and arrange the maintenance plan according to the overall health;

[0009] Step 2: Obtain the pipeline pressure, slurry flow rate and liquid level difference of the slurry in the slurry pool when the equipment is started and stopped in the historical data of the slurry circulation pump at the same power, calculate the blockage coefficient of the spray system, the spray system includes pipelines and nozzles, and calculate and determine the equipment failure rate of the desulfurization system by setting the blockage threshold and the health of the pump;

[0010] Step 3: Obtain the operating parameter data of the desulfurization system in the historical data, calculate the cost ratio of each set of data through the accumulated sulfur dioxide and the operating cost, and filter the operating parameters according to the equipment failure rate. Mark the priority of the pump history for the filtered data to form a sample data set, and then preprocess the sample data set to form a training data set;

[0011] Step 4: Build a desulfurization system operation optimization model through long short-term memory neural network, use the detection parameters in the training data set as the input of the optimization model, and the execution parameters in the training data set as the output of the optimization model to train the model, then collect real-time detection parameters as the input of the model, and output the optimization results through the optimization model;

[0012] Step 5: A group of sparrow individuals are randomly generated through the sparrow search algorithm. Each individual represents a set of hyperparameter configurations for running the optimization model. A better hyperparameter space is explored by adjusting the individual position. The optimization model is trained and run using the hyperparameter configuration of each individual. The loss function is used as the fitness evaluation indicator for optimization.

[0013] Furthermore, the specific steps of obtaining the health factor of the pump by calculating the stability and performance loss degree of the pump are:

[0014] The historical data of current, vibration, temperature and noise are counted according to a certain time period TNa, and the historical data of each day is collected once every time interval ω, and the volatility of the current, vibration, temperature and noise of the pump is calculated respectively. The stability of the pump is calculated according to the weight of the volatility of current, vibration, temperature and noise;

[0015] The fluctuation formula of the current, vibration, temperature and noise of the pump is:

[0016]

[0017] Among them, AIa is the fluctuation of current, AZa is the fluctuation of vibration, AWa is the fluctuation of temperature, ASa is the fluctuation of noise, xi ia is the current at time ia, xz ia is the vibration at time ia, xw ia is the temperature at time ia, xs ia is the noise at time ia, ia is the data collection time, Na is the number of data collection times in the time period TNa;

[0018] The calculation formula for pump stability is:

[0019] A=β1*AIa+β2*AZa+β3*AWa+β4*ASa

[0020] Among them, A is the stability of the pump, β1, β2, β3, and β4 are the weights of the volatility of current, vibration, temperature, and noise, respectively.

[0021] Furthermore, the calculation formula for the performance loss degree of the pump is:

[0022]

[0023] Among them, B is the performance loss degree of the pump, A ta is the pump stability during time period ta, A tb is the pump stability during time period tb, and ta is the pump stability during the initial operation of the desulfurization system.

[0024] Furthermore, the calculation formula of the health factor of the pump is:

[0025]

[0026] Among them, Q is the health factor of the pump, B is the degree of performance loss of the pump, and A is ta is the pump stability during time period ta, A tb is the stability of the pump during time period tb, and B is the degree of performance loss of the pump;

[0027] The specific method of judging the health of the pump according to the health coefficient and arranging the maintenance plan according to the overall health is as follows:

[0028] When Q>γ1, the pump is judged to be in a healthy state, with a priority of 1, and no maintenance is required;

[0029] When γ2≤W≤γ1, the pump is judged to be in normal state, with a priority of 2, and no maintenance is required;

[0030] When Q<γ2, the pump is judged to be in an unhealthy state, with a priority of 3, and needs to be repaired;

[0031] Among them, γ1 and γ2 are the judgment criteria for health and unhealthiness, respectively, and γ1>γ2.

[0032] Furthermore, the calculation steps of the blockage coefficient of the spray system are as follows:

[0033] The calculation formula of the pipeline blockage coefficient is:

[0034]

[0035] Wherein, C is the blockage coefficient of the pipeline, Lic is the slurry flow rate of the pipeline at the current moment, Lno is the slurry flow rate of the pipeline at the time when the maintenance is completed, Pic is the pipeline pressure of the pipeline at the current moment, and Pno is the pipeline pressure of the pipeline at the time when the maintenance is completed;

[0036] The calculation formula of the nozzle blockage coefficient is:

[0037]

[0038] Wherein, D is the blockage coefficient of the nozzle, Htd1 is the liquid level difference of the slurry pool before and after the slurry pump circulation pump is started when the nozzle is not blocked, td1 is the time when the liquid level difference reflux disappears when the nozzle is not blocked, Htd2 is the liquid level difference of the slurry pool before and after the slurry pump circulation pump is started at the current moment, and td1 is the time when the liquid level difference reflux disappears at the current moment;

[0039] The calculation formula of the blockage coefficient of the spray system is:

[0040] L=C+D

[0041] Wherein, L is the blockage coefficient of the sprinkler system;

[0042] When L≥γ3, the sprinkler system is judged to be blocked;

[0043] The calculation formula of the equipment failure rate of the desulfurization system is:

[0044]

[0045] Among them, Gz is the failure rate of the desulfurization system, Ly is the blocked spray system, Ngd is the total number of spray systems, Qy is the unhealthy pump, and Nbg is the total number of pumps.

[0046] Further, the operating parameters include detection parameters and execution parameters. The detection parameters include flue gas intake volume, flue gas temperature, sulfur dioxide concentration, slurry pool level, limestone concentration, pH value, gypsum concentration, number of circulation pumps, number of oxygen pumps, real-time power consumption unit price, real-time cost of limestone, and system cumulative desulfurization time. The execution parameters include the start and stop status of the circulation pump and the start and stop status of the oxygen pump.

[0047] The calculation formula for calculating the cost ratio of each set of data by the cumulative amount of sulfur dioxide and the operating cost is:

[0048]

[0049] Among them, V is the cost ratio, Pd is the real-time power consumption unit price, Ps is the real-time cost of limestone, Vin is the flue gas intake volume, and Nso is the sulfur dioxide concentration;

[0050] The specific method of screening the operating parameters and marking the priority of the pump history on the screened data to form a sample data set is:

[0051] The historical data is divided into days, with every ten days of data as a group. The data is sorted according to the cost ratio in descending order, and the data is sorted according to the equipment failure rate in descending order. The data in the first two tenths of the order in the two groups of data are filtered out, and the filtered data are marked with timestamps. At the same time, the circulation pumps and oxygen pumps in the data are numbered and marked with priorities.

[0052] Further, the desulfurization system operates an optimization model, including an input gate, a forget gate, and an output gate;

[0053] The equation of the forget gate is:

[0054] f t =δ(W f ·[h t―1 ,x t ]+b f )

[0055] Among them, f t is the current switch state of the forget gate, δ represents the sigmoid function, W f is the forget gate weight, b f is the forget gate bias parameter, h t―1 is the hidden state at the previous moment, x t Input switch for current data;

[0056] Determine how much of the unit state at the previous moment needs to be retained to the current moment;

[0057] The equation for the input gate is:

[0058] i t =δ(W i ·[h t―1 ,x t ]+b i )

[0059] Among them, i tis the current switch state of the input gate, δ represents the sigmoid function, W i is the input gate weight, b i is the input gate bias parameter, h t―1 is the output hidden state of the previous moment, x t is the current detection parameter data;

[0060]

[0061] in, Candidate cell state, tanh represents the hyperbolic tangent function, W c is the candidate cell weight, b c is the candidate cell state bias parameter, h t―1 is the hidden state at the previous moment, x t Input switch for current data;

[0062] Determine how much of the network's input data needs to be saved to the unit state at the current moment;

[0063] The update equation is;

[0064]

[0065] Among them, C t Current cell state, C t―1 is the cell state at the previous moment;

[0066] The equation for the output gate is:

[0067] o t =(W o ·[h t―1 ,x t ]+b o )

[0068] h t =o t *tanh(C t )

[0069] Among them, t is the current switch state of the output gate, δ represents the sigmoid function, W o is the output gate weight, b o is the output gate bias parameter, h t―1 is the hidden state at the previous moment, x t is the current data input switch, h t It is the execution parameter data at the current moment;

[0070] Controls how much of the current unit state needs to be output to the current output value;

[0071] Among them, the mathematical expressions of the sigmoid function and the hyperbolic tangent function are;

[0072]

[0073] Among them, δ(x γ ) is the sigmoid function, x γ is the input of sigmoid function, tanh(C t ) is the hyperbolic tangent function, C t Input to the function.

[0074] Further, the process of the sparrow search algorithm may include being a finder, a follower;

[0075] The finder's position update formula is:

[0076]

[0077] in, is the position of the i-th sparrow in the t-th generation in the d-th dimension, sparrow i is the weight type in the optimization model running in the hyperparameter configuration, dimension d is the weight dimension in the optimization model running in the hyperparameter configuration, Q is a random number that follows a normal distribution, R 2 is a uniform random number in [0,1], ST is the warning threshold, and its value range is [0.5,1];

[0078] Follower position update formula;

[0079]

[0080] in, is the position of the sparrow with the best fitness in the population at the current iteration number t in the dth dimension, and rand{―1,1} is a random number generated from -1 to 1;

[0081] The calculation formula of the loss function is:

[0082]

[0083] Among them, nm is the number of samples, y im is the true value, yc im is the predicted value.

[0084] The present invention also provides a system based on operation optimization of a desulfurization system, wherein the system is used to execute the above-mentioned method based on operation optimization of a desulfurization system, comprising:

[0085] A data acquisition model, which is used to obtain the historical data of the current, vibration, temperature, and noise of the slurry circulation pump and oxygen pump, the pipeline pressure, slurry flow rate, and the level difference of the slurry in the slurry pool when the equipment is started and stopped, and the operating parameter data of the desulfurization system;

[0086] An equipment analysis module, which is used to calculate the blockage coefficient and equipment failure rate of the spray system according to the pipeline pressure of the slurry circulation pump at the same power, the slurry flow rate and the liquid level difference of the slurry in the slurry pool when the equipment is started and stopped, and calculate the cost ratio of each set of data through the accumulated amount of sulfur dioxide and the operating cost;

[0087] A data screening module is used to screen the operating parameters according to the cost ratio and the equipment failure rate, mark the priority of the pump history on the screened data to form a sample data set, and then pre-process the sample data set to form a training data set

[0088] An optimization system module, wherein the optimization system module constructs a desulfurization system operation optimization model through a long short-term memory neural network, uses the detection parameters in the training data set as the input of the optimization model, and the execution parameters in the training data set as the output of the optimization model to train the model, and then collects the real-time detection parameters as the input of the model, and outputs the optimization result through the optimization model;

[0089] The system parameter optimization module is used to randomly generate a group of sparrow individuals through a sparrow search algorithm, each individual represents a set of hyperparameter configurations for running an optimization model, explore a better hyperparameter space by adjusting the individual position, use the hyperparameter configuration of each individual to train the running optimization model, and optimize it using a loss function as a fitness evaluation indicator.

[0090] Compared with the prior art, the beneficial effects of the present invention are as follows: the present invention obtains the health coefficient of the pump and judges the health degree by calculating the real-time stability and performance loss degree of the liquid circulation pump and the oxygen pump, calculates the blockage coefficient of the spray system, and calculates and judges the equipment failure rate of the desulfurization system by setting the blockage threshold and the health degree of the pump. According to the cost ratio of the desulfurization system and the equipment failure rate, the historical operation parameter data are screened to form a training data set, a desulfurization system operation optimization model is constructed through a long short-term memory neural network, and the operation optimization model is trained. Then, the real-time detection parameters are collected as the input of the model, the optimization result is output through the optimization model, and the operation optimization model is optimized through the sparrow search algorithm and the loss function as the fitness evaluation index.

[0091] The present invention screens historical operating parameter data to form a training data set based on the cost ratio of the desulfurization system and the equipment failure rate, fully considering the energy consumption of the slurry circulation pump and the oxygen pump, the cost of the desulfurizer and the factors of equipment failure, and optimizes the operation of the desulfurization system through the screened data, effectively eliminating the situation where the desulfurization effect is good but the equipment is damaged and the additional energy consumption cost and the cost of the reagent are too high, so that the desulfurization system is optimized more comprehensively and meticulously from multiple angles, and a desulfurization system operation optimization model is constructed by a long short-term memory neural network, and the optimization model is optimized, so that the overall parameters of the desulfurization system can be optimized more accurately and quickly. BRIEF DESCRIPTION OF THE DRAWINGS

[0092] Figure 1 It is a schematic diagram of the overall method flow of the present invention;

[0093] Figure 2 It is a schematic diagram of the overall system structure of the present invention. DETAILED DESCRIPTION

[0094] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0095] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0096] Example:

[0097] See also Figure 1 , the present invention provides a technical solution:

[0098] A method for optimizing the operation of a desulfurization system, comprising the following specific steps:

[0099] Step 1: Obtain the historical data of current, vibration, temperature and noise of the desulfurization system pump to be optimized, including the slurry circulation pump and oxygen pump. Obtain the health coefficient of the pump by calculating the stability and performance loss degree of the pump, judge the health of the pump based on the health coefficient, and arrange the maintenance plan based on the overall health.

[0100] The specific steps of obtaining the health factor of the pump by calculating the stability and performance loss degree of the pump are:

[0101] The historical data of current, vibration, temperature and noise are counted according to a certain time period TNa, and the historical data of each day is collected once every time interval ω. The volatility of the pump's current, vibration, temperature and noise is calculated respectively, and the stability of the pump is calculated according to the weights of the volatility of current, vibration, temperature and noise.

[0102] The main structure of the slurry circulation pump and oxygen pump is that the motor drives the impeller. Therefore, the health of the slurry circulation pump and oxygen pump can be judged by current, vibration, temperature, and noise. The magnitude of the current is directly related to the load of the pump. Under normal circumstances, the current should be kept within the design range. Abnormal fluctuations in the current may indicate changes in the load of the pump, such as blockage, wear, or motor problems. For example, an increase in current usually means an increase in the resistance of the pump, which may be caused by blockage or mechanical failure of the pump. The vibration level can reflect the mechanical stability of the pump. During normal operation, the vibration should be kept at a low level. Abnormal vibration is usually related to mechanical failures such as imbalance, looseness, or bearing damage. Increased temperature usually means increased friction, poor lubrication, or overload, which may cause component damage. Continuous high temperature can provide early warning of potential failures. The pump will generate a certain amount of noise during normal operation. Abnormal noise changes can indicate internal faults. For example, increased noise may indicate cavitation, impeller damage, or other mechanical problems. By comprehensively analyzing the volatility of these parameters, the operating status and health of the slurry circulation pump and oxygen pump can be fully identified.

[0103] In this embodiment, the fluctuation formulas of the current, vibration, temperature and noise of the pump are:

[0104]

[0105] Among them, AIa is the fluctuation of current, AZa is the fluctuation of vibration, AWa is the fluctuation of temperature, ASa is the fluctuation of noise, xi ia is the current at time ia, xz ia is the vibration at time ia, xw ia is the temperature at time ia, xs ia is the noise at time ia, ia is the data collection time, and Na is the number of data collection times in the time period TNa.

[0106] The time length of TNa is one day, and the fluctuations of the pump's current, vibration, temperature, and noise within one day are calculated.

[0107] The calculation formula for pump stability is:

[0108] A=β1*AIa+β2*AZa+β3*AWa+β4*ASa

[0109] Among them, A is the stability of the pump, β1, β2, β3, and β4 are the weights of the volatility of current, vibration, temperature, and noise, respectively.

[0110] The stability of the pump reflects the stable state of the pump. The higher the real-time stability, the greater the probability of pump damage. The real-time stability index can help identify potential failure risks. Through the changes in real-time stability, problems can be discovered in time.

[0111] In this embodiment, the calculation formula for the performance loss degree of the pump is:

[0112]

[0113] Among them, B is the performance loss degree of the pump, A ta is the pump stability during time period ta, A tb is the pump stability during time period tb, and ta is the pump stability during the initial operation of the desulfurization system.

[0114] The pump stability at the initial operation of the desulfurization system is the most representative, reflecting the best stability of the pump. The statistical time length nt of the pump stability can be one day or more days, reducing the impact of accidental factors on the judgment of pump stability.

[0115] The degree of performance loss is used to quantify the gap between the overall performance of the pump in the current state and the overall performance in the rated state. The greater the degree of performance loss, the lower the current working efficiency of the pump.

[0116] In this embodiment, the calculation formula of the health factor of the pump is:

[0117]

[0118] Among them, Q is the health factor of the pump, B is the degree of performance loss of the pump, and A is ta is the pump stability during time period ta, A tb is the stability of the pump during time period tb, and B is the degree of performance loss of the pump;

[0119] The specific method of judging the health of the pump according to the health coefficient and arranging the maintenance plan according to the overall health is as follows:

[0120] When Q>γ1, the pump is judged to be in a healthy state, with a priority of 1, and no maintenance is required;

[0121] When γ2≤Q≤γ1, the pump is judged to be in normal state, with a priority of 2, and no maintenance is required;

[0122] When Q<γ2, the pump is judged to be in an unhealthy state, with a priority of 3, and needs to be repaired;

[0123] Among them, γ1 and γ2 are the judgment criteria for health and unhealthiness, respectively, and γ1>γ2.

[0124] Step 2: Obtain the pipeline pressure, slurry flow rate and liquid level difference of the slurry in the slurry pool when the equipment is started and stopped in the historical data of the slurry circulation pump at the same power, and calculate the blockage coefficient of the spray system. The spray system includes pipelines and nozzles. By setting the blockage threshold and the health of the pump, calculate and determine the equipment failure rate of the desulfurization system.

[0125] In the desulfurization system, since the pressure in the pipeline is a key parameter in fluid dynamics, when the spray system is operating normally, the pressure in the pipeline should be maintained within a certain range. If blockage occurs, the pressure will be affected, usually manifested as an increase in pressure drop. By monitoring the pressure change, it can be determined whether there is an increase in flow resistance. Flow rate is an important indicator of liquid delivery capacity. In the spray system, the stability of the flow rate and the change in flow rate value can reflect the working efficiency and blockage of the system. When the flow rate is significantly lower than the expected value, it may indicate that there is a blockage in the system, resulting in obstruction of liquid flow. The level difference of the slurry in the slurry pool also provides important information when the equipment is started and stopped. The change in liquid level reflects the inflow and outflow of the fluid. If the liquid level changes abnormally, such as the liquid level fails to fall back quickly during shutdown, it may indicate a blockage that prevents the liquid from flowing freely.

[0126] In this embodiment, the calculation steps of the blockage coefficient of the spray system are as follows:

[0127] The calculation formula of the pipeline blockage coefficient is:

[0128]

[0129] Among them, C is the blockage coefficient of the pipeline, Lic is the slurry flow rate of the pipeline at the current moment, Lno is the slurry flow rate of the pipeline when the maintenance is completed, Pic is the pipeline pressure of the pipeline at the current moment, and Pno is the pipeline pressure of the pipeline when the maintenance is completed.

[0130] The calculation formula of the nozzle blockage coefficient is:

[0131]

[0132] Among them, D is the blockage coefficient of the nozzle, Htd1 is the liquid level difference of the slurry pool before and after the slurry pump circulation pump is started when the nozzle is not blocked, td1 is the time when the liquid level difference reflux disappears when the nozzle is not blocked, Htd2 is the liquid level difference of the slurry pool before and after the slurry pump circulation pump is started at the current moment, and td1 is the time when the liquid level difference reflux disappears at the current moment.

[0133] The calculation formula of the blockage coefficient of the spray system is:

[0134] L=C+D

[0135] Wherein, L is the blockage coefficient of the sprinkler system;

[0136] When L≥γ3, the sprinkler system is judged to be blocked.

[0137] The blockage coefficient is a comprehensive indicator. Generally speaking, an increase in the blockage coefficient means that the flow is blocked and the fluid cannot pass through the system smoothly at the designed flow rate. This coefficient can be used to link pressure loss, flow rate and liquid level changes through a formula to form a quantitative standard for monitoring and judging blockage conditions.

[0138] The calculation formula of the equipment failure rate of the desulfurization system is:

[0139]

[0140] Among them, Gz is the failure rate of the desulfurization system, Ly is the blocked spray system, Ngd is the total number of spray systems, Qy is the unhealthy pump, and Nbg is the total number of pumps.

[0141] The equipment failure rate indicates the proportion of failed equipment in the desulfurization system. The equipment failure rate means that the desulfurization system cannot be used normally or seriously affects the desulfurization effect.

[0142] Step 3: Obtain the operating parameter data of the desulfurization system from the historical data, calculate the cost ratio of each set of data through the accumulated sulfur dioxide and the operating cost, and filter the operating parameters according to the equipment failure rate. Mark the priority of the pump history for the filtered data to form a sample data set, and then preprocess the sample data set to form a training data set.

[0143] In this embodiment, the operating parameters include detection parameters and execution parameters. The detection parameters include flue gas intake volume, flue gas temperature, sulfur dioxide concentration, slurry pool level, limestone concentration, pH value, gypsum concentration, number of circulation pumps, number of oxygen pumps, real-time power consumption unit price, real-time cost of limestone, and system cumulative desulfurization time. The execution parameters include the start and stop status of the circulation pump and the start and stop status of the oxygen pump.

[0144] The flue gas intake directly affects the scale and efficiency of the desulfurization reaction. Too high an intake may result in insufficient desulfurizer limestone to react completely, reducing the desulfurization efficiency. The flue gas temperature affects the rate and efficiency of the desulfurization reaction. Higher temperatures usually accelerate the reaction, but too high a temperature may cause equipment damage or incomplete reaction. The sulfur dioxide concentration affects the desulfurizer consumption and reaction rate. Too high a concentration will accelerate the desulfurizer loss. Too low a level in the slurry pool may result in dry pumping or insufficient desulfurizer concentration, while too high a level may affect the reaction efficiency or overflow the slurry pool. The limestone concentration directly affects the efficiency of the desulfurization reaction. Insufficient concentration will result in incomplete removal of sulfur dioxide. The pH value affects the desulfurization reaction. Generally, the optimal pH range is between 5.5 and 7.5. Too high or too low will affect the reaction efficiency and gypsum formation. Too high a gypsum concentration may cause precipitation or scaling, affecting system operation. The number of circulating pumps affects the circulation efficiency of the slurry. Insufficient quantity may result in poor slurry flow, thus affecting the desulfurization effect, while too many will lead to waste of resources. The number of oxygen pumps is related to the supply of oxygen in the reaction, affecting the oxidation reaction of limestone and the removal efficiency of sulfur dioxide. Power consumption affects operating costs, which in turn affects the overall economy. High power consumption may lead to increased operating expenses. The cost of limestone directly affects the economy of the desulfurization system. Rising costs may lead to reduced operating profits. The accumulated desulfurization time is related to the operating efficiency and wear of the equipment. Long-term operation may lead to increased equipment failures.

[0145] The filter data is marked with the priority of the pump history to form a sample data set, and then the sample data set is preprocessed to form a training data set. Marking priority training for each pump helps to select according to the health status of each pump to prevent unexpected shutdowns due to pump health problems during the desulfurization process, resulting in serious consequences such as sulfur dioxide emissions not meeting standards and forced shutdowns. The preprocessing uses the normalization method to scale each data value to a range between 0 and 1, so that all input data have the same value range, avoiding unreasonable effects on the model training process due to large values ​​of certain features. Normalizing the data can improve training efficiency and model accuracy.

[0146] In this embodiment, the calculation formula for calculating the cost ratio of each set of data by using the accumulated amount of sulfur dioxide and the operating cost is:

[0147]

[0148] Among them, V is the cost ratio, Pd is the real-time power consumption unit price, Ps is the real-time cost of limestone, Vin is the flue gas intake volume, and Nso is the sulfur dioxide concentration.

[0149] The specific method of screening the operating parameters and marking the priority of the pump history on the screened data to form a sample data set is:

[0150] The historical data is divided into days, with every ten days of data as a group. The data is sorted according to the cost ratio in descending order, and the data is sorted according to the equipment failure rate in descending order. The data in the first two tenths of the order in the two groups of data are filtered out, and the filtered data are marked with timestamps. At the same time, the circulation pumps and oxygen pumps in the data are numbered and marked with priorities.

[0151] The cost ratio reflects the economic cost per unit of sulfur dioxide removed. By screening out operating parameter combinations with low cost ratios, operating expenses can be effectively reduced and economic benefits can be improved. Equipment failure rate is a key indicator for evaluating the health of the system. A high failure rate may lead to frequent downtime and maintenance, increasing operating costs. By analyzing the equipment failure rate, parameters that are stable during operation can be screened out to reduce the risk of failure and improve the overall reliability of the system. By combining cost ratio and failure rate for screening, operating parameters can be evaluated from multiple perspectives to ensure that the system is not only economically reasonable but also stable in operation.

[0152] Step 4: Build a desulfurization system operation optimization model through long short-term memory neural network, use the detection parameters in the training data set as the input of the optimization model, and the execution parameters in the training data set as the output of the optimization model to train the model, then collect real-time detection parameters as the input of the model, and output the optimization results through the optimization model.

[0153] Long short-term memory neural networks can effectively capture long-term dependencies in time series data. This is particularly important for optimizing operating parameters in desulfurization systems, because these data are usually time-related. Long short-term memory neural networks can learn the changing patterns of data in different time periods, thereby more accurately predicting the execution parameters of desulfurization systems. There is usually a complex nonlinear relationship between operating parameters and system performance in desulfurization systems. Long short-term memory neural networks can effectively model these nonlinear relationships, thereby improving the accuracy and effect of optimization.

[0154] In this embodiment, the desulfurization system operation optimization model includes an input gate, a forget gate and an output gate;

[0155] The input gate controls the input information that flows into the current unit state. The input gate it receives the current input and the hidden state of the previous moment, and uses the sigmoid function to determine how much input information to use, generating a value in the range of [0,1];

[0156] The equation for the input gate is:

[0157] i t =δ(W i ·[h t―1 ,x t ]+b i )

[0158] Among them, i t is the current switch state of the input gate, δ represents the sigmoid function, W i is the input gate weight, b i is the input gate bias parameter, h t―1 is the output hidden state of the previous moment, x t is the current detection parameter data;

[0159]

[0160] in, Candidate cell state, tanh represents the hyperbolic tangent function, W c is the candidate cell weight, b c is the candidate cell state bias parameter, h t―1 is the hidden state at the previous moment, x t Input switch for current data;

[0161] Determine how much of the network's input data needs to be saved to the unit state at the current moment;

[0162] The update equation is;

[0163]

[0164] Among them, C t Current cell state, C t―1 is the cell state at the previous moment;

[0165] The forget gate controls the historical information that flows into the current unit state. For neurons in the hidden layer of the long short-term memory neural network, some historical information must be forgotten before operation. The forget gate receives the hidden state of the previous moment and the input data input switch of the current moment, which is mapped to the range of [0,1] by the sigmoid function.

[0166] The equation of the forget gate is:

[0167] f t =δ(W f ·[h t―1 ,x t ]+b f )

[0168] Among them, f t is the current switch state of the forget gate, δ represents the sigmoid function, Wf is the forget gate weight, b f is the forget gate bias parameter, h t―1 is the hidden state at the previous moment, x t Input switch for current data;

[0169] Determine how much of the unit state at the previous moment needs to be retained to the current moment;

[0170] The output gate controls the influence of the current memory unit on the current execution parameter data, that is, which part will be output at the current moment. The output gate receives the hidden state of the previous moment and the current moment data input switch, which is mapped to the range of [0,1] by the sigmoid function;

[0171] The equation for the output gate is:

[0172] o t =(W o ·[h t―1 ,x t ]+b o )

[0173] h t =o t *tanh(C t )

[0174] Among them, t is the current switch state of the output gate, δ represents the sigmoid function, W o is the output gate weight, b o is the output gate bias parameter, h t―1 is the hidden state at the previous moment, x t is the current data input switch, h t It is the execution parameter data at the current moment;

[0175] Controls how much of the current unit state needs to be output to the current output value;

[0176] Among them, the mathematical expressions of the sigmoid function and the hyperbolic tangent function are;

[0177]

[0178] Among them, δ(x γ ) is the sigmoid function, x γ is the input of sigmoid function, tanh(C t ) is the hyperbolic tangent function, C t Input to the function.

[0179] Step 5: A group of sparrow individuals are randomly generated through the sparrow search algorithm. Each individual represents a set of hyperparameter configurations for running the optimization model. A better hyperparameter space is explored by adjusting the individual position. The optimization model is trained and run using the hyperparameter configuration of each individual. The loss function is used as the fitness evaluation indicator for optimization.

[0180] The sparrow search algorithm is a swarm intelligence optimization algorithm that can effectively explore the parameter space and avoid falling into the local optimal solution. This is crucial for the hyperparameter tuning of the long short-term memory neural network model, because the appropriate hyperparameter setting can significantly improve the performance of the model. The sparrow search algorithm can accelerate the search process and improve the convergence speed by simulating the foraging behavior of sparrows. This enables the long short-term memory neural network to find the best parameter combination faster during training, thereby shortening the training time. The sparrow search algorithm shows good adaptability in dynamic environments. For the long short-term memory neural network model that needs to be adjusted in real time, it can dynamically optimize the hyperparameters according to the latest data to improve the flexibility and stability of the model.

[0181] In this embodiment, the process of the sparrow search algorithm may include being a finder and a follower;

[0182] The finder's position update formula is:

[0183]

[0184] in, is the position of the i-th sparrow in the t-th generation in the d-th dimension, sparrow i is the weight type in the optimization model running in the hyperparameter configuration, dimension d is the weight dimension in the optimization model running in the hyperparameter configuration, Q is a random number that follows a normal distribution, R 2 is a uniform random number in [0,1], ST is the warning threshold, and its value range is [0.5,1].

[0185] Follower position update formula;

[0186]

[0187] in, is the position of the sparrow with the best fitness in the population at the current iteration number t in the dth dimension, and rand{―1,1} is a random number generated from -1 to 1.

[0188] Use the validation set to calculate the performance indicators (such as accuracy, loss function, etc.) of each LSTM model as the fitness value;

[0189] The calculation formula of the loss function is:

[0190]

[0191] Among them, nm is the number of samples, y im is the true value, yc im is the predicted value.

[0192] See also Figure 2 The present invention also provides a system based on operation optimization of a desulfurization system, wherein the system is used to execute the above-mentioned method based on operation optimization of a desulfurization system, comprising:

[0193] A data acquisition model, which is used to obtain the historical data of the current, vibration, temperature, and noise of the slurry circulation pump and oxygen pump, the pipeline pressure, slurry flow rate, and the level difference of the slurry in the slurry pool when the equipment is started and stopped, and the operating parameter data of the desulfurization system;

[0194] An equipment analysis module, which is used to calculate the blockage coefficient and equipment failure rate of the spray system according to the pipeline pressure of the slurry circulation pump at the same power, the slurry flow rate and the liquid level difference of the slurry in the slurry pool when the equipment is started and stopped, and calculate the cost ratio of each set of data through the accumulated amount of sulfur dioxide and the operating cost;

[0195] A data screening module is used to screen the operating parameters according to the cost ratio and the equipment failure rate, mark the priority of the pump history on the screened data to form a sample data set, and then pre-process the sample data set to form a training data set

[0196] An optimization system module, wherein the optimization system module constructs a desulfurization system operation optimization model through a long short-term memory neural network, uses the detection parameters in the training data set as the input of the optimization model, and the execution parameters in the training data set as the output of the optimization model to train the model, and then collects the real-time detection parameters as the input of the model, and outputs the optimization result through the optimization model;

[0197] The system parameter optimization module is used to randomly generate a group of sparrow individuals through a sparrow search algorithm, each individual represents a set of hyperparameter configurations for running an optimization model, explore a better hyperparameter space by adjusting the individual position, use the hyperparameter configuration of each individual to train the running optimization model, and optimize it using a loss function as a fitness evaluation indicator.

[0198] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0199] The above embodiments may be implemented in whole or in part by software, hardware, firmware or any other combination thereof. When implemented by software, the above embodiments may be implemented in whole or in part in the form of a computer program product. Those skilled in the art may appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein may be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0200] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, and may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0201] The above description is only a specific implementation manner of the present application, but the protection scope of the present application is not limited thereto. Any technician familiar with the technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application.

Claims

1. A method for optimizing the operation of a desulfurization system, characterized in that: The specific steps include: Step 1: Obtain the historical data of current, vibration, temperature and noise of the desulfurization system pump to be optimized, including the slurry circulation pump and the oxygen pump, calculate the stability and performance loss degree of the pump, obtain the health factor of the pump, judge the health of the pump according to the health factor, and arrange the maintenance plan according to the overall health; Step 2: Obtain the pipeline pressure, slurry flow rate and liquid level difference of the slurry in the slurry pool when the equipment is started and stopped in the historical data of the slurry circulation pump at the same power, calculate the blockage coefficient of the spray system, the spray system includes pipelines and nozzles, and calculate and determine the equipment failure rate of the desulfurization system by setting the blockage threshold and the health of the pump; Step 3: Obtain the operating parameter data of the desulfurization system in the historical data, calculate the cost ratio of each set of data through the accumulated sulfur dioxide and the operating cost, and filter the operating parameters according to the equipment failure rate. Mark the priority of the pump history for the filtered data to form a sample data set, and then preprocess the sample data set to form a training data set; Step 4: Build a desulfurization system operation optimization model through long short-term memory neural network, use the detection parameters in the training data set as the input of the optimization model, and the execution parameters in the training data set as the output of the optimization model to train the model, then collect real-time detection parameters as the input of the model, and output the optimization results through the optimization model; Step 5: A group of sparrow individuals are randomly generated through the sparrow search algorithm. Each individual represents a set of hyperparameter configurations for running the optimization model. A better hyperparameter space is explored by adjusting the individual position. The optimization model is trained and run using the hyperparameter configuration of each individual. The loss function is used as the fitness evaluation indicator for optimization.

2. The method for optimizing the operation of a desulfurization system according to claim 1, characterized in that: The specific steps of obtaining the health factor of the pump by calculating the stability and performance loss degree of the pump are: The historical data of current, vibration, temperature and noise are counted according to a certain time period TNa, and the historical data of each day is collected once every time interval ω, and the volatility of the current, vibration, temperature and noise of the pump is calculated respectively. The stability of the pump is calculated according to the weight of the volatility of current, vibration, temperature and noise; The fluctuation formula of the current, vibration, temperature and noise of the pump is: Among them, AIa is the fluctuation of current, AZa is the fluctuation of vibration, AWa is the fluctuation of temperature, ASa is the fluctuation of noise, xi ia is the current at time ia, xz ia is the vibration at time ia, xw ia is the temperature at time ia, xs ia is the noise at time ia, ia is the data collection time, Na is the number of data collection times in the time period TNa; The calculation formula for pump stability is: A=β1*AIa+β2*AZa+β3*AWa+β4*ASa Among them, A is the stability of the pump, β1, β2, β3, and β4 are the weights of the volatility of current, vibration, temperature, and noise, respectively.

3. The method for optimizing the operation of a desulfurization system according to claim 2, characterized in that: The calculation formula for the performance loss degree of the pump is: Among them, B is the performance loss degree of the pump, A ta is the pump stability during time period ta, A tb is the pump stability during time period tb, and ta is the pump stability during the initial operation of the desulfurization system.

4. The method for optimizing the operation of a desulfurization system according to claim 3, characterized in that: The calculation formula of the health factor of the pump is: Among them, Q is the health factor of the pump, B is the degree of performance loss of the pump, and A is ta is the pump stability during time period ta, A tb is the stability of the pump during time period tb, and B is the degree of performance loss of the pump; The specific method of judging the health of the pump according to the health coefficient and arranging the maintenance plan according to the overall health is as follows: When Q>γ1, the pump is judged to be in a healthy state, with a priority of 1, and no maintenance is required; When γ2≤Q≤γ1, the pump is judged to be in normal state, with a priority of 2, and no maintenance is required; When Q < γ2, the pump is judged to be in an unhealthy state, with a priority of 3, and needs to be repaired; Among them, γ1 and γ2 are the judgment criteria of health and unhealthiness respectively, and γ1>γ2.

5. The method for optimizing the operation of a desulfurization system according to claim 1, characterized in that: The calculation steps of the blockage coefficient of the spray system are as follows: The calculation formula of the pipeline blockage coefficient is: Wherein, C is the blockage coefficient of the pipeline, Lic is the slurry flow rate of the pipeline at the current moment, Lno is the slurry flow rate of the pipeline at the time when the maintenance is completed, Pic is the pipeline pressure of the pipeline at the current moment, and Pno is the pipeline pressure of the pipeline at the time when the maintenance is completed; The calculation formula of the nozzle blockage coefficient is: Wherein, D is the blockage coefficient of the nozzle, Htd1 is the liquid level difference of the slurry pool before and after the slurry pump circulation pump is started when the nozzle is not blocked, td1 is the time when the liquid level difference reflux disappears when the nozzle is not blocked, Htd2 is the liquid level difference of the slurry pool before and after the slurry pump circulation pump is started at the current moment, and td1 is the time when the liquid level difference reflux disappears at the current moment; The calculation formula of the blockage coefficient of the spray system is L = C + D Wherein, L is the blockage coefficient of the sprinkler system; When L≥γ3, the sprinkler system is judged to be blocked; The calculation formula of the equipment failure rate of the desulfurization system is: Among them, Gz is the failure rate of the desulfurization system, Ly is the blocked spray system, Ngd is the total number of spray systems, Qy is the unhealthy pump, and Nbg is the total number of pumps.

6. The method for optimizing the operation of a desulfurization system according to claim 1, characterized in that: The operating parameters include detection parameters and execution parameters. The detection parameters include flue gas intake volume, flue gas temperature, sulfur dioxide concentration, slurry pool level, limestone concentration, pH value, gypsum concentration, number of circulation pumps, number of oxygen pumps, real-time power consumption unit price, real-time limestone cost, and system cumulative desulfurization time. The execution parameters include the start and stop status of the circulation pump and the start and stop status of the oxygen pump. The calculation formula for calculating the cost ratio of each set of data by the cumulative amount of sulfur dioxide and the operating cost is: Among them, V is the cost ratio, Pd is the real-time power consumption unit price, Ps is the real-time cost of limestone, Vin is the flue gas intake volume, and Nso is the sulfur dioxide concentration; The specific method of screening the operating parameters and marking the priority of the pump history on the screened data to form a sample data set is: The historical data is divided into days, with every ten days of data as a group. The data is sorted according to the cost ratio in descending order, and the data is sorted according to the equipment failure rate in descending order. The data in the first two tenths of the order in the two groups of data are filtered out, and the filtered data are marked with timestamps. At the same time, the circulation pumps and oxygen pumps in the data are numbered and marked with priorities.

7. The method for optimizing the operation of a desulfurization system according to claim 1, characterized in that: The desulfurization system runs an optimization model, including an input gate, a forget gate, and an output gate; The equation of the forget gate is: f t =δ(W f ·[h t-1 ,x t ]+b f ) Among them, f t is the current switch state of the forget gate, δ represents the sigmoid function, W f is the forget gate weight, b f is the forget gate bias parameter, h t-1 is the hidden state at the previous moment, x t Input switch for current data; Determine how much of the unit state at the previous moment needs to be retained to the current moment; The equation for the input gate is: i t =δ(W i ·[h t-1 ,x t ]+b i ) Among them, i t is the current switch state of the input gate, δ represents the sigmoid function, W i is the input gate weight, b i is the input gate bias parameter, h t-1 is the output hidden state of the previous moment, x t is the current detection parameter data; in, Candidate cell state, tanh represents the hyperbolic tangent function, W c is the candidate cell weight, b c is the candidate cell state bias parameter, h t-1 is the hidden state at the previous moment, x t Input switch for current data; Determine how much of the network's input data needs to be saved to the unit state at the current moment; The update equation is; Among them, C t Current cell state, C t-1 is the cell state at the previous moment; The equation for the output gate is: o t =(W o ·[h t-1 ,x t ]+b o ) h t =o t *tanh(C t ) Among them, t is the current switch state of the output gate, δ represents the sigmoid function, W o is the output gate weight, b o is the output gate bias parameter, h t-1 is the hidden state at the previous moment, x t is the current data input switch, h t It is the execution parameter data at the current moment; Controls how much of the current unit state needs to be output to the current output value; Among them, the mathematical expressions of the sigmoid function and the hyperbolic tangent function are; Among them, δ(x γ ) is the sigmoid function, x γ is the input of sigmoid function, tanh(C t ) is the hyperbolic tangent function, C t Input to the function.

8. The method for optimizing the operation of a desulfurization system according to claim 1, characterized in that: The process of the sparrow search algorithm includes a finder and a follower; The finder's position update formula is: in, is the position of the i-th sparrow in the t-th generation in the d-dimensional update, sparrow i is the weight type in the optimization model running in the hyperparameter configuration, dimension d is the weight dimension in the optimization model running in the hyperparameter configuration, Q is a random number that follows a normal distribution, R2 is a uniform random number in [0,1], ST is the warning threshold, and the value range is [0.5,1]; Follower position update formula; in, is the position of the sparrow with the best fitness in the population at the current iteration number t in the dth dimension, and rand{―1,1} is a random number generated from -1 to 1; The calculation formula of the loss function is: Among them, nm is the number of samples, y im is the true value, yc im is the predicted value.

9. A system based on operation optimization of a desulfurization system, characterized in that: The operation optimization system is used to execute the method for operation optimization based on the desulfurization system according to any one of claims 1 to 8, comprising: A data acquisition model, which is used to obtain the historical data of the current, vibration, temperature, and noise of the slurry circulation pump and oxygen pump, the pipeline pressure, slurry flow rate, and the level difference of the slurry in the slurry pool when the equipment is started and stopped, and the operating parameter data of the desulfurization system; An equipment analysis module, which is used to calculate the blockage coefficient and equipment failure rate of the spray system according to the pipeline pressure of the slurry circulation pump at the same power, the slurry flow rate and the liquid level difference of the slurry in the slurry pool when the equipment is started and stopped, and calculate the cost ratio of each set of data through the accumulated amount of sulfur dioxide and the operating cost; A data screening module is used to screen the operating parameters according to the cost ratio and the equipment failure rate, mark the priority of the pump history on the screened data to form a sample data set, and then pre-process the sample data set to form a training data set An optimization system module, wherein the optimization system module constructs a desulfurization system operation optimization model through a long short-term memory neural network, uses the detection parameters in the training data set as the input of the optimization model, and the execution parameters in the training data set as the output of the optimization model to train the model, and then collects the real-time detection parameters as the input of the model, and outputs the optimization result through the optimization model; The system parameter optimization module is used to randomly generate a group of sparrow individuals through a sparrow search algorithm, each individual represents a set of hyperparameter configurations for running an optimization model, explore a better hyperparameter space by adjusting the individual position, use the hyperparameter configuration of each individual to train the running optimization model, and optimize it using a loss function as a fitness evaluation indicator.

Images