AI-based boiler operation control system optimization method and system

Abstract

The invention provides an AI-based boiler operation control system optimization method and system, and relates to the technical field of boiler operation control systems.The method comprises the steps that historical steam plan demand quantity, steam actual demand quantity and environment data are obtained, fault data of steam using equipment and timestamps printed on production duration are obtained, and the steam using equipment is optimized; establishing a steam demand prediction model based on a long short-term memory neural network; acquiring the steam prediction set at the current moment as the input of the trained model; predicting the actual steam demand in the future; establishing a boiler steam regulation model based on a fuzzy neural network; a comprehensive cost function is constructed by acquiring steam adjustment data and production data when steam quantity adjustment is performed on the boiler every time, and parameters in the fuzzy neural network are updated according to a gradient descent method by taking the comprehensive cost function as a target.

Description

Technical Field

[0001] The present invention relates to the technical field of boiler operation control systems, and specifically to an optimization method and system for a boiler operation control system based on AI. Background Art

[0002] In modern manufacturing, steam, as one of the important energy sources, is crucial for the normal operation of production lines, especially in industrial production processes such as cigarette factories. The stability of steam supply directly affects production efficiency and product quality, especially when the steam demand fluctuates greatly and the ambient temperature changes frequently. Traditional steam supply control systems often struggle to respond to environmental changes and sudden failures of production equipment in real time, resulting in increased volatility of steam supply, thereby affecting production stability and product quality. Although some existing intelligent control methods and technologies can optimize energy management to a certain extent, due to the lack of effective prediction models and adjustment mechanisms, the problem of energy supply stability has not been fully solved.

[0003] Traditional boiler steam control systems mostly rely on manual experience or fixed parameter adjustment, and are difficult to adapt to complex working conditions with multiple factors coupled such as ambient temperature fluctuations, equipment failure rates, and dynamic production demands during the production process. Existing cigarette factory systems have insufficient research on the dynamic correlation between steam demand prediction and equipment status and environmental variables, resulting in a large deviation between the planned demand and the actual demand. The end adjustment lacks an intelligent closed-loop mechanism, and the fuel input and cooling adjustment parameters rely on empirical adjustment, with a lag in response and prone to causing energy waste or supply fluctuations. These problems lead to insufficient stability of steam pressure and flow, which may trigger quality risks such as excessive moisture content of cut tobacco and process parameter drift, and there is an urgent need to build an optimization control system integrating data-driven and intelligent decision-making.

[0004] The above information disclosed in the background art section is only used to enhance the understanding of the background of the present disclosure, and thus it may include information that does not constitute the prior art known to those of ordinary skill in the art. Summary of the Invention

[0005] The purpose of the present invention is to provide an optimization method and system for a boiler operation control system based on AI to solve the problems raised in the above background art.

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

[0007] An optimization method for a boiler operation control system based on AI, the specific steps include:

[0008] Step 1: Obtain the historical steam planned demand, actual steam demand, and environmental data of the steam-using equipment of the boiler to be optimized, as well as the failure data and production duration of the steam-using equipment. Calculate the failure rate based on the failure data and production time. Stamp the steam planned demand, environmental data, and failure rate with timestamps and correspond them one by one. After preprocessing, form a steam prediction set;

[0009] Step 2: Establish a steam demand prediction model based on a long short-term memory neural network. Use the steam prediction set of the previous moment as the input of the model, and the actual steam demand of the current moment as the output of the model to train the model. Obtain the steam prediction set of the current moment as the input of the trained model, and predict the actual steam demand in the future;

[0010] Step 3: Obtain the historical adjustment parameters of the boiler. Establish a boiler steam adjustment model based on a fuzzy neural network. Use the steam deviation value between the actual demands of the previous moment and the current moment and the environmental data to form an input data set as the input of the model, and the historical adjustment parameters as the output to train the model. Calculate the steam deviation value based on the actual steam demand of the current moment and the predicted actual steam demand in the future as the input of the model, and obtain the adjustment parameters;

[0011] Step 4: Obtain the historical coal price. Calculate the adjustment rate cost by calculating the steam adjustment rate according to the steam adjustment data when the boiler adjusts the steam volume. Calculate the production loss caused by insufficient steam according to the production data. Construct a comprehensive cost function based on the adjustment rate cost and the production loss cost;

[0012] Step 5: Take the comprehensive cost function as the objective, optimize the parameters of the fuzzy neural network, update the parameters in the fuzzy neural network according to the gradient descent method. When the change amount of the cost function is less than the preset cost threshold, stop the optimization and obtain the optimized parameters of the fuzzy neural network.

[0013] Further, the failure data includes the number of failures, failure duration, and failure time;

[0014] The calculation steps of the failure rate are as follows:

[0015] Segment the failure data at the same time interval to form time windows. Calculate the failure rate for the failure data within the window: where the time windows are formed by the time intervals, and the failure rate is calculated for the failure data within the window:

[0016]

[0017] where is the failure data within the window, are the number of failures, failure duration, and failure time respectively;

[0018] The calculation formula for the failure rate is as follows:

[0019]

[0020] Wherein, is the failure rate, is the production duration within the window, is the th failure duration within the window, is the number of failures within the window.

[0021] Furthermore, the calculation formula for the preprocessing is:

[0022]

[0023] Wherein, is the steam prediction set, is the data after preprocessing, , are respectively the minimum and maximum values in the steam prediction set.

[0024] Furthermore, the steam demand prediction model established based on the long short-term memory neural network consists of 3 gates, namely the forget gate, the input gate, and the output gate:

[0025] The equation expression of the forget gate is:

[0026]

[0027] Wherein, is the current switch state of the forget gate, represents the sigmoid function, is the weight of the forget gate, is the bias parameter of the forget gate, is the hidden state at the previous moment;

[0028] determines how much of the cell state at the previous moment needs to be retained to the current moment;

[0029] The equation expression of the input gate is:

[0030]

[0031] Wherein, is the current switch state of the input gate, is the weight of the input gate, is the bias parameter of the input gate, is the output hidden state at the previous moment, is the steam prediction set at the previous moment:

[0032]

[0033] Among them, The candidate cell state Represents the hyperbolic tangent function Is the candidate cell weight Is the candidate cell state bias parameter;

[0034] Determines how much of the input data of the network at the current moment needs to be saved to the cell state;

[0035] The update equation is:

[0036]

[0037] Among them, The current cell state Is the cell state at the previous moment;

[0038] The equation expression of the output gate is:

[0039]

[0040]

[0041] Among them, Is the current switch state of the output gate Is the output gate weight Is the output gate bias parameter Is the actual steam demand at the current moment;

[0042] Controls how much of the current cell state needs to be output to the actual steam demand at the current moment.

[0043] Furthermore, the environmental data includes temperature and humidity;

[0044] The adjustment parameters include the amount of coal added and the amount of cooling water sprayed;

[0045] The boiler steam adjustment model established based on the fuzzy neural network consists of an input layer, a fuzzy layer, a calculation layer, and an output layer. The specific logic is:

[0046] The first layer is the input layer, representing the input parameters. This layer has Nodes Represents the Input data in the input dataset, and the formula is:

[0047]

[0048] Among them, Is the input input dataset;

[0049] The second layer is the fuzzification layer. The Gaussian function is used as the membership function in this layer to calculate the membership degree of the input variable as follows:

[0050]

[0051] where is the membership degree of the input variable , and are the central value and width value of the fuzzy set respectively, , , is the number of fuzzy classification levels for the input;

[0052] The third layer is the fuzzy rule calculation layer, which can perform normalization operations on the fuzzified parameters. The formula is:

[0053]

[0054] where is the product of the membership degrees of the input parameters;

[0055] The fourth layer is the output layer, which completes the center average defuzzification operation and outputs the result. The formula is:

[0056]

[0057] where is the adjustment parameter, is the fuzzy system parameter.

[0058] Furthermore, the steam regulation data includes: the steam quantity generated in real time and the coal quantity added in real time;

[0059] The steps for calculating the steam regulation rate are as follows: sample the real-time steam quantity at the same time interval , judge the steam regulation node and steam regulation type by calculating the steam quantity difference based on the steam quantities of adjacent samples, and calculate the regulation rate and regulation rate cost according to the steam regulation data within the steam regulation node;

[0060] The formula for the steam quantity difference is:

[0061]

[0062] where is the steam quantity generated by the boiler at time, is the steam quantity difference at time;

[0063] When When it is in a certain state, the adjustment type is positive adjustment;

[0064] When it is in another state, the adjustment type is negative adjustment;

[0065] When it changes from positive to negative, it is a negative adjustment node;

[0066] When it changes from negative to positive, it is a positive adjustment node;

[0067] The calculation formula for the steam adjustment rate is:

[0068]

[0069] Wherein, is the steam adjustment rate, , are the steam flows of two adjacent positive and negative adjustment nodes, is the time between two adjacent adjustment nodes.

[0070] Furthermore, the calculation formula for the adjustment rate cost is:

[0071]

[0072] Wherein, , is the amount of coal added at time, is the coal price added at time;

[0073] Production losses caused by insufficient steam:

[0074]

[0075] Wherein, , is the amount of coal added at time, is the coal price added at time.

[0076] Furthermore, the calculation formula for the comprehensive cost function is:

[0077]

[0078] Wherein, is the comprehensive cost function.

[0079] Furthermore, the steps to update the parameters in the fuzzy neural network according to the gradient descent method are:

[0080] Calculate the gradient of the comprehensive cost function:

[0081]

[0082] where are the parameters of the fuzzy neural network, are the center values respectively when 、width values , is the number of time steps, , is the production loss cost caused by insufficient steam, is the comprehensive cost function;

[0083] The update formula is:

[0084]

[0085] where is the learning rate, are the updated parameters, are the unupdated parameters;

[0086] When the change amount of the cost function is less than the preset cost threshold, stop the optimization:

[0087]

[0088] where is the preset cost threshold.

[0089] The present invention also provides an optimization method and system for a boiler operation control system based on AI. The optimization system for the boiler operation control system for the intelligent acquisition system of mine data is used to execute the above-mentioned optimization method for the boiler operation control system, including:

[0090] A prediction data acquisition module, which is used to obtain the historical steam planned demand, steam actual demand and environmental data of the steam using equipment of the boiler to be optimized, as well as the fault data and production duration of the steam using equipment, calculate the failure rate according to the fault data and production time, stamp the steam planned demand, environmental data and failure rate with time stamps and correspond them one by one, and form a steam prediction set after preprocessing;

[0091] A steam prediction module, which is used to establish a steam demand prediction model based on a long short-term memory neural network, use the steam prediction set of the previous moment as the input of the model, use the steam actual demand of the current moment as the output of the model to train the model, obtain the steam prediction set of the current moment as the input of the trained model, and predict the future steam actual demand;

[0092] The boiler regulation module is used to establish a boiler steam regulation model based on a fuzzy neural network. The steam deviation value between the actual demand at the previous moment and the current moment and the environmental data form an input data set, which is used as the input of the model. The historical regulation parameters are used as the output to train the model, and the steam deviation value calculated based on the actual steam demand at the current moment and the predicted steam demand in the future is used as the input of the model to obtain the regulation parameters.

[0093] The optimized data module is used to obtain the historical coal price, calculate the regulation rate cost according to the steam regulation data when the boiler adjusts the steam volume, calculate the production loss caused by insufficient steam according to the production data, and construct a comprehensive cost function based on the regulation rate cost and the production loss cost.

[0094] The optimized regulation module is used to optimize the parameters of the fuzzy neural network with the comprehensive cost function as the target, update the parameters in the fuzzy neural network according to the gradient descent method, and stop the optimization when the change amount of the cost function is less than the preset cost threshold to obtain the optimized parameters of the fuzzy neural network.

[0095] Compared with the prior art, the beneficial effects of the present invention are as follows: The present invention obtains the historical planned steam demand, actual steam demand and environmental data, as well as the fault data and production duration of the steam using equipment, stamps time stamps, forms a steam prediction set after preprocessing, establishes a steam demand prediction model based on a long short-term memory neural network, obtains the steam prediction set at the current moment as the input of the trained model, predicts the actual steam demand in the future, establishes a boiler steam regulation model based on a fuzzy neural network, obtains the regulation parameters, constructs a comprehensive cost function by obtaining the steam regulation data and production data each time the boiler adjusts the steam volume, and updates the parameters in the fuzzy neural network according to the gradient descent method with the comprehensive cost function as the target.

[0096] The present invention can significantly improve the stability and accuracy of steam supply. Firstly, the steam demand prediction model can effectively utilize historical data and environmental information to predict changes in steam demand in advance, avoiding problems such as steam shortages or surpluses caused by sudden demand fluctuations. By establishing a time-series correlation model based on equipment failure rates, environmental temperatures, and historical demand data, the accuracy of steam demand prediction is significantly improved, solving the problem of poor adaptability of traditional static models to sudden working conditions. Secondly, the fuzzy neural network can dynamically adjust fuel input and cooling parameters when regulating the boiler steam output, thus ensuring the matching of steam supply and production demand and reducing the additional costs caused by over-regulation or regulation delays. In addition, by comprehensively considering the adjustment rate cost and production loss cost, this solution can reduce energy waste and production losses while optimizing the adjustment, thereby improving the economy and efficiency of the entire steam supply system. At the same time, the optimization process based on the gradient descent method further refines the adjustment strategy of the fuzzy neural network, enabling the system to always maintain a stable operating state in a changing production environment and effectively avoiding quality risks caused by steam supply fluctuations. BRIEF DESCRIPTION OF THE DRAWINGS

[0097] Figure 1 is a schematic diagram of the overall method flow of the present invention;

[0098] Figure 2 is a schematic diagram of the overall system structure of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0099] To make the objectives, technical solutions, and advantages of the present invention clearer and more understandable, the present invention will be further described in detail below in conjunction with specific embodiments.

[0100] It should be noted that unless otherwise defined, the technical terms or scientific terms used in the present invention should have the ordinary meanings understood by those of ordinary skill in the field to which the present invention belongs. The "first", "second", and similar terms used in the present invention do not indicate any order, quantity, or importance, but are only used to distinguish different components. The terms "including" or "comprising" and the like mean that the elements or objects appearing before this term cover the elements or objects listed after this term and their equivalents, without excluding other elements or objects. The terms "connected" or "coupled" and the like are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. The terms "upper", "lower", "left", "right", etc. are only used to represent relative positional relationships, and when the absolute position of the object being described changes, the relative positional relationship may also change accordingly.

[0101] Embodiment:

[0102] Please refer to Figure 1 , the present invention provides a technical solution:

[0103] An optimization method for a boiler operation control system based on AI, the specific steps include:

[0104] Step 1: Obtain the historical steam planned demand, actual steam demand, and environmental data of the steam-using equipment of the boiler to be optimized, as well as the fault data and production duration of the steam-using equipment. Calculate the failure rate based on the fault data and production time. Stamp the steam planned demand, environmental data, and failure rate with timestamps and correspond them one by one. After preprocessing, form a steam prediction set.

[0105] The steam planned demand represents the predetermined operating demand of the boiler and is usually predicted based on the production plan, equipment load, and requirements of the steam-using equipment. By analyzing the historical planned demand, a long short-term memory neural network can identify the relationship between the planned demand and the actual demand, helping to optimize the supply and use of steam.

[0106] The ambient temperature directly affects the efficiency of boiler operation and the demand for steam production. High-temperature weather may reduce the load demand of the boiler, while low-temperature weather may increase the steam demand. A long short-term memory neural network can model the relationship between the ambient temperature and the steam demand, thereby improving the prediction accuracy.

[0107] The failure rate of the equipment is an important factor affecting the steam production volume. Equipment failures may cause production interruptions or efficiency reductions, thereby affecting the steam demand. Through the historical failure rate data, a long short-term memory neural network can learn the patterns of failure occurrences and adjust the steam demand prediction based on the likelihood of failures.

[0108] In this embodiment, the fault data includes the number of faults, fault duration, and fault time;

[0109] The calculation steps of the failure rate are as follows:

[0110] At the same time interval Segment the fault data, A time window is formed by [number of] time intervals, and the failure rate is calculated for the fault data within the window:

[0111]

[0112] Wherein, is the fault data within the window, are the number of faults, fault duration, and fault time respectively;

[0113] The calculation formula for the failure rate is:

[0114]

[0115] Wherein, is the failure rate, is the production duration within the window, is the th failure duration within the window, is the number of failures within the window.

[0116] Traditional static failure rate calculation uses the global average value, which cannot reflect the changing pattern of equipment status over time. By dividing a fixed time window, the sliding window method calculates the ratio of failure time to production duration within the window, enabling the calculation result to dynamically track the equipment performance degradation cycle. For example, when a device enters a high-failure period after continuous high-load operation, the sliding window can promptly capture this trend, while the static method may mask the current risk due to the dilution effect of historical data. Using the sliding window method to calculate the failure rate can also be well combined with timestamps, synchronizing data such as real-time ambient temperature and steam demand with the failure data. Each time the failure rate is calculated, it is synchronously updated using the latest real-time data and the corresponding timestamps, thus ensuring that the entire failure rate calculation reflects both historical trends and closely combines with the current actual operating status.

[0117] In this embodiment, the calculation formula for the preprocessing is:

[0118]

[0119] where, is the steam prediction set, is the data after preprocessing, , are respectively the minimum and maximum values in the steam prediction set.

[0120] Maximum normalization, also known as max normalization, is a commonly used data preprocessing method. Its basic operation is to scale the values of the data proportionally to a specified range, usually [0, 1], for subsequent machine learning or neural network processing.

[0121] The boiler steam demand prediction model needs to process various feature data, such as steam demand, ambient temperature, failure rate, etc. There may be significant differences in the dimensions and numerical ranges of these features (for example, the temperature is in the tens of degrees, and the steam demand may be several tons). Directly using the original data as input to the model may cause some features to have too much influence on model training and some features to be ignored. Maximum normalization maps all features to the same range, making the contributions of each feature to the model roughly the same and avoiding the weights of some features being overly amplified.

[0122] Step 2: Establish a steam demand prediction model based on the long short-term memory neural network. Use the steam prediction set at the previous moment as the input of the model, and the actual steam demand at the current moment as the output of the model to train the model. Obtain the steam prediction set at the current moment as the input of the trained model, and predict the actual steam demand in the future.

[0123] The steam demand of the boiler system is usually time series data, affected by various factors (such as historical steam demand, environmental temperature, equipment failures, etc.). The demand of steam-using equipment has significant time dependence. The steam demand shows periodic fluctuations over time (such as seasonal changes, production cycles, etc.), and is also affected by unexpected events (such as equipment failures, temperature changes, etc.).

[0124] Since the long short-term memory neural network, as a deep learning model specialized in processing time series data, can well handle the long-term dependencies and dynamic changes in time series data, and can learn the trends, seasonal fluctuations, and sudden abnormal changes in the data. Therefore, in boiler operation control, the long short-term memory neural network can effectively capture these long-term dependencies through its special gating structure (such as forget gate, input gate, output gate), analyze the complex relationships between different time nodes, especially the seasonal fluctuations of steam demand, the demand changes caused by equipment failures, and the impact of the external environment on demand. This enables the long short-term memory neural network to comprehensively consider factors such as historical demand trends, environmental temperature changes, and equipment failures during the prediction process, thereby improving the accuracy of prediction.

[0125] In this embodiment, the steam demand prediction model established based on the long short-term memory neural network consists of 3 gates, namely the forget gate, the input gate, and the output gate:

[0126] The equation expression of the forget gate is:

[0127]

[0128] Where, is the current switch state of the forget gate, represents the sigmoid function, is the weight of the forget gate, is the bias parameter of the forget gate, is the hidden state at the previous moment;

[0129] determines how much of the cell state at the previous moment needs to be retained to the current moment;

[0130] The equation expression of the input gate is:

[0131]

[0132] Where, is the current switch state of the input gate, is the weight of the input gate, is the bias parameter of the input gate, is the output hidden state at the previous moment, is the steam prediction set at the previous moment:

[0133]

[0134] Among them, candidate cell state, represents the hyperbolic tangent function, is the candidate cell weight, is the bias parameter of the candidate cell state;

[0135] determines how much of the input data of the network at the current moment needs to be saved to the cell state;

[0136] The update equation is:

[0137]

[0138] Among them, current cell state, is the cell state at the previous moment;

[0139] The equation expression of the output gate is:

[0140]

[0141]

[0142] Among them, is the current switch state of the output gate, is the weight of the output gate, is the bias parameter of the output gate, is the actual steam demand at the current moment;

[0143] Controls how much of the current cell state needs to be output to the actual steam demand at the current moment.

[0144] Step 3: Obtain the historical adjustment parameters of the boiler, establish a boiler steam adjustment model based on the fuzzy neural network, form an input data set with the steam deviation value between the actual demands at the previous moment and the current moment and the environmental data as the input of the model, use the historical adjustment parameters as the output to train the model, and calculate the steam deviation value as the input of the model according to the actual steam demands at the current moment and the future prediction, and obtain the adjustment parameters.

[0145] The boiler steam demand and regulation process involves many uncertainties and fuzzy factors, such as fluctuations in environmental temperature, changes in steam demand, and the operating conditions of the boiler itself. Traditional control methods may be difficult to accurately handle these complex changes, while fuzzy neural networks can process these uncertainties through fuzzy logic. For example, input data (such as steam demand deviation values and environmental data) often has imprecise or fuzzy properties, and fuzzy neural networks can make reasonable inferences under unclear circumstances, thereby improving the control effect.

[0146] Fuzzy neural networks combine the learning ability of neural networks with the reasoning ability of fuzzy logic and can learn complex non-linear relationships. The steam regulation process of the boiler may be non-linear, that is, the relationship between fuel input and cooling regulation parameters and steam demand is complex and cannot be reduced to a linear model. Through training, the network structure can be automatically adjusted to capture the complex non-linear relationships between various factors in the steam regulation process, thereby achieving more accurate control. Fuzzy neural networks can quickly make inferences based on real-time input data (such as current steam volume, environmental changes, etc.) and output regulation parameters to ensure that the difference between the steam output of the boiler and the actual demand is minimized. This real-time adaptive ability can effectively respond to various changes in the production process, reduce system fluctuations, and improve the operating efficiency and stability of the boiler.

[0147] In this embodiment, the environmental data includes temperature and humidity;

[0148] The regulation parameters include the amount of coal added and the spraying amount of cooling water;

[0149] The boiler steam regulation model established based on the fuzzy neural network consists of an input layer, a fuzzy layer, a calculation layer, and an output layer. The specific logic is as follows;

[0150] The first layer is the input layer, which represents the input parameters. There are nodes representing input data in the input data set. The formula is:

[0151]

[0152] where, is the input data set;

[0153] The second layer is the fuzzification layer. This layer uses the Gaussian function as the membership function to calculate the membership degree of the input variable . The formula is:

[0154]

[0155] where, is the input variable Membership degree of and are the central value and width value of the fuzzy set respectively, , , is the number of fuzzy classifications for the input;

[0156] The third layer is the fuzzy rule calculation layer, which can perform normalization operations on the fuzzified parameters. The formula is:

[0157]

[0158] where is the product of the membership degrees of the input parameters;

[0159] The fourth layer is the output layer, which completes the center-average defuzzification operation and outputs the result. The formula is:

[0160]

[0161] where is the adjustment parameter, is the fuzzy system parameter.

[0162] Step 4: Obtain historical coal prices, calculate the adjustment rate cost by calculating the steam adjustment rate based on the steam adjustment data when adjusting the steam flow according to the boiler, calculate the production loss caused by insufficient steam according to the production data, and construct a comprehensive cost function based on the adjustment rate cost and the production loss cost.

[0163] In this embodiment, the steam adjustment data includes: the steam flow generated in real time and the coal amount added in real time;

[0164] The steps for calculating the steam adjustment rate are as follows: Sampling the real-time steam flow at the same time interval Judging the steam adjustment node and the steam adjustment type by calculating the steam flow difference based on the steam flow of adjacent samplings, and calculating the adjustment rate and the adjustment rate cost according to the steam adjustment data within the steam adjustment node;

[0165] The calculation formula for the steam flow difference is:

[0166]

[0167] where is the steam flow generated by the boiler at time, is the steam flow difference at time;

[0168] When , the adjustment type is positive adjustment;

[0169] When it is the case, the regulation type is negative regulation;

[0170] When it changes from positive to negative, it is a negative regulation node;

[0171] When it changes from negative to positive, it is a positive regulation node;

[0172] The calculation formula for the steam regulation rate is:

[0173]

[0174] Wherein, is the steam regulation rate, , are the steam flows of two adjacent positive and negative regulation nodes, is the time of two adjacent regulation nodes;

[0175] In this embodiment, the calculation formula for the regulation rate cost is:

[0176]

[0177] Wherein, , is the amount of coal added at the moment of , is the coal price added at the moment of ;

[0178] Production losses caused by insufficient steam:

[0179]

[0180] Wherein, , is the amount of coal added at the moment of , is the coal price added at the moment of ;

[0181] In this embodiment, the calculation formula for the comprehensive cost function is:

[0182]

[0183] Wherein, is the comprehensive cost function.

[0184] The adjustment rate cost refers to the consumption of the system, energy, equipment, etc. during the boiler steam adjustment process. When the steam volume is adjusted too quickly, the boiler may consume more energy or exacerbate equipment wear, resulting in additional maintenance costs. If the adjustment is too slow, it may lead to a lag in steam supply and fail to meet production demands in a timely manner, affecting production efficiency. Optimizing the adjustment rate cost aims to balance the speed of steam adjustment so that the adjustment is neither too rapid to cause waste nor too slow to affect production efficiency. Therefore, a reasonable adjustment rate can ensure the maximization of the energy utilization rate of boiler operation, avoiding unnecessary energy waste and equipment failures.

[0185] The production loss cost refers to the losses caused to production when the steam supply is insufficient or excessive. If the boiler steam supply is insufficient, production may stagnate or the production rate may decrease, resulting in economic losses. Excessive steam will cause energy waste and increase additional costs. By calculating the production loss cost, the system can evaluate the negative impacts when the steam supply is insufficient, optimize the adjustment strategy to ensure the stability and sufficiency of steam supply, avoid production stagnation caused by insufficient steam, and thus reduce production losses.

[0186] By optimizing the adjustment rate cost, it can ensure that the boiler is both efficient and stable when adjusting the steam volume, avoiding energy waste and equipment wear caused by too fast or too slow adjustment. A reasonable adjustment rate can help save energy, reduce unnecessary maintenance expenditures, and lower the overall cost of boiler operation. At the same time, optimizing the production loss cost can ensure that the steam supply always matches the production demand, avoiding production stoppages, losses, or efficiency reduction caused by insufficient steam. Through reasonable optimization of steam adjustment, the boiler can more accurately meet the steam demand during the production process, avoiding production line stagnation or output fluctuations caused by improper adjustment. Precise adjustment reduces production uncertainty, thereby improving production efficiency and stability.

[0187] Step 5: Taking the comprehensive cost function as the objective, optimize the parameters of the fuzzy neural network, update the parameters in the fuzzy neural network according to the gradient descent method, and stop the optimization when the change amount of the cost function is less than the preset cost threshold to obtain the optimized parameters of the fuzzy neural network.

[0188] The comprehensive cost function combines the adjustment rate cost and the production loss cost, aiming to balance energy consumption, equipment load, production demand, and efficiency. It can comprehensively consider all aspects during the steam adjustment process, thus avoiding local optima in the optimization and finding a globally optimal adjustment strategy.

[0189] The gradient descent method is a commonly used optimization algorithm that can effectively minimize the objective function (comprehensive cost function). Its basic principle is to calculate the gradient of the objective function with respect to each parameter, and then adjust the parameters along the opposite direction of the gradient to gradually reduce the objective function.

[0190] The gradient is the partial derivative of the objective function with respect to the network parameters, which represents the rate of change of the objective function as the parameters change. In this case, the gradient represents the rate of change of the comprehensive cost function with respect to the parameters of the fuzzy neural network. By calculating the contribution of each parameter to the comprehensive cost function, it can be determined which parameters need to be adjusted and by how much. Continuing to perform iterative updates through the gradient descent method until the change in the cost function is less than a preset threshold. This means that the system has found a relatively stable optimization point, the parameters of the model converge, and no large-scale adjustments are made anymore.

[0191] In this process, the goal of each update is to minimize the cost function, that is, to reduce the adjustment rate cost and the production loss cost. The gradient descent method can ensure that the system finds the optimal adjustment parameters in multiple dimensions, and through backpropagation and continuous adjustment, finally achieves a better steam adjustment strategy.

[0192] With the optimized fuzzy neural network, the system can more precisely control the rate of steam adjustment, avoiding energy waste or production losses caused by over-fast or over-slow adjustment. For example, if the adjustment rate is too high, unnecessary energy consumption may occur in the boiler; while if the adjustment is too slow, it may lead to insufficient steam supply, affecting the operation of the production line. The optimized network can precisely control the adjustment rate and reduce ineffective fluctuations. The optimized neural network can smooth the adjustment process, reduce violent fluctuations, thereby extending the service life of the equipment and reducing the maintenance cost.

[0193] In this embodiment, the steps for updating the parameters in the fuzzy neural network according to the gradient descent method are as follows:

[0194] Calculate the gradient of the comprehensive cost function:

[0195]

[0196] Among them, are the parameters of the fuzzy neural network, are the center values respectively when 、width values , is the number of time steps, , is the production loss cost caused by insufficient steam, is the comprehensive cost function;

[0197] The update formula is:

[0198]

[0199] Among them, is the learning rate, is the updated parameter, is the parameter before update;

[0200] When the change in the cost function is less than a preset cost threshold, stop the optimization:

[0201]

[0202] wherein, is the preset cost threshold.

[0203] Please refer to Figure 2 , the present invention further provides an optimization method and system for a boiler operation control system based on AI. The boiler operation control system optimization system is used to execute the above-mentioned boiler operation control system optimization method, including:

[0204] A prediction data acquisition module, which is used to obtain the historical steam planned demand, actual steam demand and environmental data of the steam using equipment of the boiler to be optimized, as well as the failure data and production duration of the steam using equipment, calculate the failure rate according to the failure data and production time, stamp the steam planned demand, environmental data and failure rate with time stamps and correspond them one by one, and form a steam prediction set after preprocessing;

[0205] A steam prediction module, which is used to establish a steam demand prediction model based on a long short-term memory neural network, use the steam prediction set of the previous moment as the input of the model, and the actual steam demand of the current moment as the output of the model to train the model, obtain the steam prediction set of the current moment as the input of the trained model, and predict the actual steam demand in the future;

[0206] A boiler adjustment module, which is used to establish a boiler steam adjustment model according to a fuzzy neural network, use the steam deviation value between the actual demands of the previous moment and the current moment and environmental data to form an input data set as the input of the model, and the historical adjustment parameters as the output to train the model, and use the steam deviation value calculated according to the actual steam demand of the current moment and the predicted future steam demand as the input of the model to obtain the adjustment parameters;

[0207] An optimization data module, which is used to obtain the historical coal price, calculate the adjustment rate cost according to the steam adjustment data when the boiler adjusts the steam volume, calculate the production loss caused by insufficient steam according to the production data, and construct a comprehensive cost function according to the adjustment rate cost and production loss cost;

[0208] An optimization adjustment module, which is used to optimize the parameters of the fuzzy neural network with the comprehensive cost function as the target, update the parameters in the fuzzy neural network according to the gradient descent method, and stop the optimization when the change in the cost function is less than a preset cost threshold to obtain the optimized parameters of the fuzzy neural network.

[0209] The above formulas are all dimensionless and only take their numerical values for calculation. The formulas are obtained by collecting a large amount of data for software simulation to get a formula closest to the actual situation. The preset parameters in the formulas are set by those skilled in the art according to the actual situation.

[0210] The above embodiments can be implemented in whole or in part by software, hardware, firmware, or any combination thereof. When implemented using software, the above embodiments can be implemented in whole or in part in the form of a computer program product. Those skilled in the art can realize that the units and algorithm steps of each example described in combination with the embodiments disclosed herein can be implemented by electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are executed by hardware or software methods depends on the specific application and design constraints of the technical solution.

[0211] The units described as separate components may or may not be physically separated. The components shown as units may or may not be physical units. They can be located in one place or distributed to multiple network units. Some or all of the units can be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0212] As described above, the above is only the specific implementation manner of this application, but the protection scope of this application is not limited thereto. Any person skilled in the art can easily think of changes or substitutions within the technical scope disclosed in this application, and all of them should be covered by the protection scope of this application.

Claims

1. An optimization method for a boiler operation control system based on AI, characterized in that, The specific steps include: Step 1: Obtain the historical steam planned demand, actual steam demand, and environmental data of the steam-using equipment of the boiler to be optimized, as well as the fault data and production duration of the steam-using equipment. Calculate the failure rate based on the fault data and production time. Stamp the steam planned demand, environmental data, and failure rate with timestamps and correspond them one by one. After preprocessing, form a steam prediction set. Step 2: Establish a steam demand prediction model based on a long short-term memory neural network. Use the steam prediction set of the previous moment as the input of the model, and the actual steam demand of the current moment as the output of the model to train the model. Obtain the steam prediction set of the current moment as the input of the trained model, and predict the actual steam demand in the future. Step 3: Obtain the historical adjustment parameters of the boiler. Establish a boiler steam adjustment model based on a fuzzy neural network. Use the steam deviation value between the actual demands of the previous moment and the current moment and the environmental data to form an input data set as the input of the model, and the historical adjustment parameters as the output to train the model. Calculate the steam deviation value based on the actual steam demand of the current moment and the predicted actual steam demand in the future as the input of the model, and obtain the adjustment parameters. Step 4: Obtain the historical coal price. Calculate the adjustment rate cost by calculating the steam adjustment rate according to the steam adjustment data when the boiler adjusts the steam volume. Calculate the production loss caused by insufficient steam according to the production data. Construct a comprehensive cost function based on the adjustment rate cost and production loss cost. Step 5: Take the comprehensive cost function as the objective, optimize the parameters of the fuzzy neural network, update the parameters in the fuzzy neural network according to the gradient descent method. When the change amount of the cost function is less than the preset cost threshold, stop the optimization and obtain the optimized parameters of the fuzzy neural network.

2. The optimization method of a boiler operation control system based on AI according to claim 1, characterized in that: The fault data includes the number of faults, fault duration, and fault time. The calculation steps of the failure rate are: At the same time interval Split the fault data A time interval forms a time window, and calculate the failure rate for the fault data within the window: ， Among them, is the fault data within the window, are the number of faults, fault duration, and fault time respectively; The calculation formula of the failure rate is: ， Among them, is the failure rate, is the production duration within the window, is the th failure duration within the window, is the number of failures within the window.

3. An optimization method for a boiler operation control system based on AI according to claim 1, characterized in that: The calculation formula of the preprocessing is: ， Among them, is the steam prediction set, is the preprocessed data, , are respectively the minimum and maximum values in the steam prediction set.

4. The optimization method of a boiler operation control system based on AI according to claim 1, characterized in that: The steam demand prediction model established based on the long short-term memory neural network consists of 3 gates, namely the forget gate, input gate, and output gate: The equation expression of the forget gate is: ， Among them, is the current switch state of the forget gate, represents the sigmoid function, is the weight of the forget gate, is the bias parameter of the forget gate, is the hidden state at the previous moment; Determine how much of the unit state of the previous moment needs to be retained to the current moment. The equation expression of the input gate is: ， Among them, is the current switch state of the input gate, is the weight of the input gate, is the bias parameter of the input gate, is the output hidden state at the previous moment, is the steam prediction set at the previous moment; ， Among them, Candidate cell state, represents the hyperbolic tangent function, is the candidate cell weight, is the candidate cell state bias parameter; Determine how much of the input data of the network at the current moment needs to be saved to the unit state. The update equation is: ， Among them, the current cell state, is the cell state at the previous moment; The equation expression of the output gate is: ， ， Among them, is the current on / off state of the output gate, is the weight of the output gate, is the bias parameter of the output gate, is the actual steam demand at the current moment; Control how much of the current unit state needs to be output to the actual steam demand at the current moment.

5. A method for optimizing a boiler operation control system based on AI according to claim 1, characterized in that: The environmental data includes temperature and humidity. The adjustment parameters include the amount of coal added and the amount of cooling water sprayed. The boiler steam adjustment model established based on the fuzzy neural network consists of an input layer, a fuzzy layer, a calculation layer, and an output layer. The specific logic is: The first layer is the input layer, representing input parameters. This layer has nodes representing the input data in the input dataset. The formula is: ， Among them, is the input data set of the input; The second layer is the fuzzification layer. This layer uses the Gaussian function as the membership function to calculate the membership degree of the input variable . The formula is as follows: ， wherein, is the membership degree of the input variable , and are the central value and the width value of the fuzzy set respectively, , , is the number of fuzzy classifications for the input; The third layer is the fuzzy rule calculation layer, which can perform a normalization operation on the fuzzified parameters. The formula is: ， Among them, is the membership degree continuous product of input parameters; The fourth layer is the output layer, which completes the center average defuzzification operation and outputs the result. The formula is: ， Among them, is an adjustment parameter, is a fuzzy system parameter.

6. An optimization method for a boiler operation control system based on AI according to claim 1, characterized in that: The steam adjustment data includes: the steam volume generated in real time, the amount of coal added in real time. The steps for calculating the steam regulation rate are as follows: At the same time intervals sample the real-time steam volume, judge the steam regulation nodes and steam regulation types by calculating the steam volume difference based on the steam volumes of adjacent samples, and calculate the regulation rate and regulation rate cost according to the steam regulation data within the steam regulation nodes; The calculation formula of the steam volume difference is: ， Among them, is the amount of steam generated by the boiler at moment, is the difference in steam amount at moment; When the adjustment type is positive adjustment; When the adjustment type is negative adjustment; When changes from positive to negative, is a negative regulatory node; When changes from negative to positive, it is a positive regulatory node; The calculation formula of the steam adjustment rate is: ， Among them, is the steam regulation rate, , is the steam flow rates of two adjacent positive and negative regulation nodes, is the time of two adjacent regulation nodes.

7. An optimization method for a boiler operation control system based on AI according to claim 6, characterized in that: The calculation formula for the adjustment rate cost is as follows: ， Among them, , is the amount of coal added at time, is the coal price added at time; Production losses caused by insufficient steam: ， Among them, is the production loss cost caused by insufficient steam, is the relationship coefficient between the loss and steam, is the planned steam demand, is the actual steam demand.

8. An optimization method for a boiler operation control system based on AI according to claim 7, characterized in that: The calculation formula for the comprehensive cost function is as follows: ， Among them, is the comprehensive cost function.

9. The optimization method of a boiler operation control system based on AI according to claim 1, characterized in that: The steps for updating the parameters in the fuzzy neural network according to the gradient descent method are as follows: Calculate the gradient of the comprehensive cost function: ， Among them, are the parameters of the fuzzy neural network, are the center values respectively when and the width values , is the number of time steps, , is the production loss cost caused by insufficient steam, is the comprehensive cost function; The update formula is as follows: ， wherein, is the learning rate, is the updated parameter, is the parameter not updated; When the change amount of the cost function is less than the preset cost threshold, stop the optimization: ， wherein, is a preset cost threshold.

10. An AI-based optimization system for boiler operation control system, characterized in that: The boiler operation control system optimization system is used to execute the AI-based boiler operation control system optimization method according to any one of claims 1-9, including; A prediction data acquisition module, which is used to obtain the historical steam planned demand, actual steam demand and environmental data of the steam using equipment of the boiler to be optimized, as well as the failure data and production duration of the steam using equipment, calculate the failure rate according to the failure data and production time, timestamp the steam planned demand, environmental data and failure rate and correspond them one by one, and form a steam prediction set after preprocessing; A steam prediction module, which is used to establish a steam demand prediction model based on a long short-term memory neural network, use the steam prediction set of the previous moment as the input of the model, and the actual steam demand of the current moment as the output of the model to train the model, obtain the steam prediction set of the current moment as the input of the trained model, and predict the actual steam demand in the future; A boiler adjustment module, which is used to establish a boiler steam adjustment model according to a fuzzy neural network, use the steam deviation value between the actual demands of the previous moment and the current moment and environmental data to form an input data set as the input of the model, and the historical adjustment parameters as the output to train the model, and use the steam deviation value calculated according to the actual steam demand of the current moment and the predicted future steam demand as the input of the model to obtain the adjustment parameters; An optimization data module, which is used to obtain the historical coal price, calculate the adjustment rate cost by calculating the steam adjustment rate according to the steam adjustment data when the boiler adjusts the steam volume, calculate the production losses caused by insufficient steam according to the production data, and construct a comprehensive cost function according to the adjustment rate cost and the production loss cost; An optimization adjustment module, which is used to optimize the parameters of the fuzzy neural network with the comprehensive cost function as the target, update the parameters in the fuzzy neural network according to the gradient descent method, and stop the optimization when the change amount of the cost function is less than the preset cost threshold to obtain the optimized parameters of the fuzzy neural network.

Images