Automatic control system and method for hot water boiler

Abstract

The invention relates to the field of intelligent control, and provides an automatic control system and method for a hot water boiler, and the method comprises the steps: firstly collecting water temperature data in real time to construct a time step data set of the water temperature data, extracting current water temperature information from the time step data set, and obtaining target water temperature data set by a user from a background database; then, the real-time deviation value between the current water temperature and the target water temperature is calculated, the deviation value is input into a PID controller to generate a control instruction comprising a burner firepower adjusting value, the proportionality coefficient, the differential coefficient and the integral coefficient of the controller are dynamically determined according to the time step data set of the water temperature data, and finally, the control instruction comprising the burner firepower adjusting value is generated. And the burner is accurately controlled according to the control instruction, so that the real-time adjustment of the water temperature is realized. Therefore, the water temperature of the hot water boiler can be automatically and intelligently adjusted and controlled.

Description

Technical Field

[0001] The present application relates to the field of intelligent control, and more specifically, to an automatic control system and method for a hot water boiler. Background Art

[0002] Hot water boilers are indispensable equipment in industrial and civil fields, widely used to provide a stable supply of hot water or steam. Hot water boilers produce the required hot water or steam by burning fuel (such as natural gas, coal or oil) or using electrical energy to heat water. Efficient, safe and reliable operation is essential to ensure effective use of energy and environmental protection.

[0003] In the combustion process of hot water boilers, the control of water temperature has a great impact on the performance, efficiency, safety and maintenance of the system. The correct selection and precise adjustment of the PID control algorithm are the key to achieving stable and efficient temperature control. If the PID parameters (proportional P, integral I, differential D) are set improperly, it may cause over-regulation or under-regulation problems, thus affecting the accuracy and stability of temperature control.

[0004] Traditional PID controllers use fixed P, I, and D parameters that are determined through debugging under specific conditions. However, in actual operation, the system's operating conditions, such as load fluctuations and temperature changes, may change significantly, making the originally optimized fixed parameters no longer applicable, resulting in reduced control accuracy and slow response. In addition, when faced with nonlinear or time-varying systems, this static setting is difficult to cope with the nonlinear characteristics of the system and external interference, and is prone to over-regulation or under-regulation problems.

[0005] Therefore, a hot water boiler automation control solution is desired. Summary of the invention

[0006] In view of the shortcomings of the prior art, the present application provides an automatic control system and method for a hot water boiler.

[0007] According to one aspect of the present application, a hot water boiler automatic control method is provided, which includes: real-time acquisition of water temperature data to obtain a time step data set of water temperature data; extracting current water temperature data from the time step data set of water temperature data; extracting target water temperature data set by a user from a background database; calculating a real-time water temperature deviation value between the target water temperature data and the current water temperature data; inputting the real-time water temperature deviation value into a PID controller to obtain a control instruction, the control instruction including a burner fire power adjustment value, wherein a proportional coefficient, a differential coefficient and an integral coefficient of the PID controller are determined based on the time step data set of the water temperature data; controlling the burner based on the control instruction; Among them, the proportional coefficient, differential coefficient and integral coefficient of the PID controller are determined based on the time step data set of the water temperature data, including: time step segmentation of the time step data set of the water temperature data to obtain a sequence of local time step data sets of water temperature data; multi-scale time series implicit encoding of the sequence of local time step data sets of water temperature data to obtain a sequence of multi-scale time series implicit correlation features of water temperature data; dynamic aggregation of water temperature data time series features of the sequence of multi-scale time series implicit correlation features of water temperature data to obtain water temperature data time series dynamic aggregation features; based on the water temperature data time series dynamic aggregation features, the proportional coefficient, differential coefficient and integral coefficient of the PID controller are obtained.

[0008] According to another aspect of the present application, a hot water boiler automation control system is provided, which is used to execute the hot water boiler automation control method as described above, and includes: a water temperature time series data segmentation module, which is used to perform time step segmentation on the time step data set of the water temperature data to obtain a sequence of local time step data sets of water temperature data; a water temperature time series data encoding module, which is used to perform multi-scale time series implicit encoding on the sequence of local time step data sets of water temperature data to obtain a sequence of multi-scale time series implicit correlation features of water temperature data; a water temperature time series feature dynamic aggregation module, which is used to perform water temperature data time series feature dynamic aggregation on the sequence of multi-scale time series implicit correlation features of water temperature data to obtain water temperature data time series dynamic aggregation features; a PID controller coefficient generation module, which is used to obtain the proportional coefficient, differential coefficient and integral coefficient of the PID controller based on the water temperature data time series dynamic aggregation features.

[0009] The present application has significant technical effects due to the adoption of the above technical solutions: the hot water boiler automatic control system and method provided by the present application first collects water temperature data in real time to construct a time step data set of water temperature data, and extracts the current water temperature information therefrom, and simultaneously obtains the target water temperature data set set by the user from the background database, and then calculates the real-time deviation value between the current water temperature and the target water temperature, and inputs this deviation value into the PID controller to generate a control instruction including the burner firepower adjustment value, the proportional coefficient, differential coefficient and integral coefficient of the controller are dynamically determined according to the time step data set of the water temperature data, and finally, the burner is precisely controlled according to this control instruction to realize the real-time adjustment of the water temperature. In this way, the automatic and intelligent adjustment and control of the water temperature of the hot water boiler can be realized. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0011] Figure 1 Flow chart of a hot water boiler automatic control method according to an embodiment of the present application.

[0012] Figure 2 Flow chart of step S150 in the hot water boiler automatic control method according to an embodiment of the present application.

[0013] Figure 3 This is a flowchart of step S153 in the hot water boiler automatic control method according to an embodiment of the present application.

[0014] Figure 4 4 is a system block diagram of a hot water boiler automatic control system according to an embodiment of the present application. DETAILED DESCRIPTION

[0015] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0016] As a core equipment in the industrial and civil fields, hot water boilers play an important role in providing a stable supply of hot water and steam. Its operating principle is to heat water to the required temperature by burning various fuels (such as natural gas, coal or fuel oil) or using electricity to generate hot water or steam. Ensuring the efficient, safe and reliable operation of the boiler plays a vital role in promoting the effective use of energy and environmental protection.

[0017] In the combustion process of hot water boilers, the precise control of water temperature has a profound impact on the overall performance, energy efficiency, safety performance and maintenance of the system. The correct selection and precise adjustment of the PID control algorithm are the key to achieving stable and efficient control of water temperature. If the PID parameters (i.e., proportional P, integral I, differential D) are improperly configured, it may lead to over-regulation or under-regulation, thereby weakening the accuracy and stability of temperature control.

[0018] Traditional PID controllers usually rely on preset P, I, and D parameters for operation, which are obtained through debugging under specific working conditions. However, in actual operation, the working conditions faced by the system (such as load fluctuations, temperature changes, etc.) often change significantly, making the originally optimized fixed parameters no longer applicable, resulting in a decline in control accuracy and a slow response speed. Especially when facing systems with nonlinear or time-varying characteristics, this static parameter setting is difficult to effectively deal with the nonlinear characteristics within the system and external interference, which can easily lead to problems of over-regulation or under-regulation.

[0019] Based on this, the present application proposes a hot water boiler automatic control method. Figure 1 Flow chart of the hot water boiler automatic control method according to the embodiment of the present application. Figure 1 As shown, according to the hot water boiler automation control method of the embodiment of the present application, the method includes: S110, real-time collection of water temperature data to obtain a time step data set of water temperature data; S120, extracting current water temperature data from the time step data set of water temperature data; S130, extracting target water temperature data set by the user from the background database; S140, calculating the real-time deviation value of water temperature between the target water temperature data and the current water temperature data; S150, inputting the real-time deviation value of water temperature into a PID controller to obtain a control instruction, the control instruction including a burner firepower adjustment value, wherein the proportional coefficient, differential coefficient and integral coefficient of the PID controller are determined based on the time step data set of the water temperature data; S160, controlling the burner based on the control instruction.

[0020] In step S110, water temperature data is collected in real time to obtain a time step data set of water temperature data. It should be understood that the hot water boiler system will be dynamically affected by various factors, such as sudden changes in load, fluctuations in ambient temperature, etc. In order to adapt to the dynamically changing system environment, so that the control system can respond quickly according to the latest water temperature situation, it is necessary to collect water temperature data in real time. The time step data set of water temperature data obtained by collecting water temperature data in real time can directly reflect the actual temperature of hot water in the hot water boiler at different time points. For temperature control, only by timely understanding the current water temperature state can the deviation from the target water temperature be accurately calculated, and then a reasonable control instruction is issued through the PID controller. It should be noted that the time step data set of water temperature data collected in real time provides information on the dynamic changes of the system, which can provide a data basis for the PID controller to adjust the proportional coefficient, integral coefficient and differential coefficient according to the actual situation, so as to effectively cope with the time-varying characteristics of the system, thereby overcoming the defects of the traditional PID controller.

[0021] The process of collecting water temperature data in real time to build a time-step data set of water temperature data involves multiple key steps and technical points to ensure that accurate and representative water temperature data can be obtained, which can provide a solid data foundation for subsequent automated control. The following is a detailed description of its implementation process: In the hot water boiler system, the reasonable deployment of water temperature sensors is the key to obtaining water temperature data. These sensors are distributed in multiple key locations of the boiler, such as inside the water tank, the inlet and outlet of the hot water pipe, etc., aiming to comprehensively and accurately capture the real-time changes in water temperature. There are many types of sensors, among which thermocouple sensors and thermistor sensors are more common. Thermocouple sensors work based on the thermoelectric effect. Their advantages are that they have a wide measurement range and can adapt to the needs of water temperature measurement in high temperature environments. They also have high measurement accuracy, which provides a strong guarantee for the accuracy of water temperature data. Thermistor sensors, with their highly sensitive characteristics to temperature changes, can quickly respond to subtle fluctuations in water temperature, and are particularly suitable for scenarios with high requirements for water temperature change accuracy. Its working principle is to accurately measure water temperature through the change in resistance value caused by water temperature changes, ensuring that the system can capture instantaneous changes in water temperature in a timely manner.

[0022] In order to ensure the timeliness and effectiveness of the collected water temperature data, the sensor continuously monitors and collects the water temperature at carefully set time intervals. The determination of the time interval is not arbitrary, and it requires comprehensive consideration of multiple factors such as the operating characteristics of the hot water boiler, the control accuracy requirements, and the rational use of system resources. In the field of industrial hot water boilers, due to the high requirements for water temperature stability and accuracy in the production process, the time interval is usually set to a short range of seconds to minutes.

[0023] The water temperature data collected by the sensor is not used directly for control, but needs to go through a series of complex transmission and processing processes. In the process of data transmission, in order to prevent data loss or errors, a variety of advanced data verification and error correction technologies are used. By adding check bits and other means, the transmitted data is strictly verified to ensure the integrity and accuracy of the data during the transmission process. Once a data error is detected, the data acquisition module will quickly start the corresponding error correction mechanism, such as requesting the sensor to retransmit data or using built-in algorithms to repair data to ensure data reliability.

[0024] When the data is successfully transmitted to the data acquisition module, the module will carefully organize and store the water temperature data in chronological order, thereby constructing a time step data set of water temperature data. Each data point in this data set carries the true state of the water temperature in the boiler at a specific time point. For example, in a system set to a 1-minute acquisition interval, the data set may present a series of continuous water temperature records, such as 80°C at 10:00, 80.3°C at 10:01, and 80.5°C at 10:02. These seemingly simple data points not only intuitively reflect the current value of the water temperature, but also contain information on the trend of the water temperature over time. Through in-depth analysis of these data, the system can gain insight into the changing laws of water temperature and predict its future trends, thereby providing a key basis for subsequent precise control.

[0025] In step S120, the current water temperature data is extracted from the time step data set of the water temperature data. It should be understood that the hot water boiler system is a dynamic system and the water temperature is changing all the time. In order to understand the real-time status of the hot water boiler system, it is necessary to extract the current water temperature data from the time step data set of the water temperature data. By continuously extracting the current water temperature data and comparing it with the target water temperature in real time, it can be determined whether the water temperature is too high, too low or close to the target value. Based on this, it is determined what kind of adjustment measures the burner should take, such as increasing the firepower, reducing the firepower or maintaining the status quo, thereby realizing dynamic control of the hot water boiler, thereby ensuring efficient, safe and reliable operation of the hot water boiler.

[0026] In step S130, the target water temperature data set by the user is extracted from the background database. It should be understood that the target water temperature data set by the user is the desired state of the hot water boiler operation. Extracting this data from the background database can clarify the target value of the hot water boiler water temperature control. The system will continuously adjust the firepower of the burner according to the target water temperature so that the water temperature can stably approach the target value, which helps to achieve accurate control of the hot water boiler water temperature.

[0027] In step S140, the real-time deviation value of the water temperature between the target water temperature data and the current water temperature data is calculated. It should be understood that the real-time deviation value of the water temperature obtained by calculation can intuitively reflect the gap between the current water temperature and the target water temperature, and it clearly contains the control direction information of whether the water temperature needs to be increased or decreased. Control based on this deviation value can achieve efficient water temperature regulation. Specifically, the system can select a suitable adjustment strategy according to the size and change trend of the deviation value to avoid over-regulation or under-regulation. For example, when the deviation value gradually decreases, indicating that the water temperature is approaching the target water temperature, the control system can correspondingly gradually reduce the adjustment range of the burner firepower so that the water temperature can smoothly reach the target water temperature to reduce the overshoot phenomenon of the water temperature, thereby improving the efficiency of temperature regulation.

[0028] In step S150, the real-time water temperature deviation value is input into a PID controller to obtain a control instruction, wherein the control instruction includes a burner power adjustment value, wherein the proportional coefficient, differential coefficient and integral coefficient of the PID controller are determined based on the time step data set of the water temperature data.

[0029] It should be understood that by inputting the real-time water temperature deviation value into the PID controller, the controller can adjust the parameters in real time according to the current water temperature deviation, thereby generating control instructions that adapt to the dynamic changes of the system. The control instruction specifically refers to the burner firepower adjustment value. In detail, when the water temperature deviation is large, the controller will output a larger firepower adjustment value to enable the burner to heat the water quickly; when the water temperature is close to the target water temperature and the deviation is small, the controller will output a smaller adjustment value to achieve fine temperature control. Through this dynamic, deviation-based control method, the accuracy of water temperature control can be improved, water temperature fluctuations can be reduced, energy efficiency can be improved, and the system's adaptability to external factors such as load changes and environmental interference can be enhanced.

[0030] Accordingly, in the determination of the proportional coefficient, differential coefficient and integral coefficient of the PID controller based on the time step data set of the water temperature data, the technical concept of the present application is to use artificial intelligence-based data processing and analysis technology to perform time step segmentation and multi-scale time series association coding on the water temperature data, so as to intelligently generate the proportional coefficient, differential coefficient and integral coefficient of the PID controller according to the representation obtained by performing time series dynamic aggregation on the sequence characteristics of the multi-scale time series implicit association characteristics of the water temperature data by multi-scale time series implicit coding of the sequence of the local time step data set of the water temperature data. In this way, the P, I, and D parameters of the PID controller can be adjusted in real time to ensure that it is always in the optimal state, and good control performance can be maintained even when the working conditions change. And it can capture the nonlinear behavior and complex patterns of the system, so that the controller can automatically adapt to new dynamic situations, avoiding the problem that traditional fixed parameters cannot cope with changes.

[0031] Specifically, Figure 2 FIG. 1 is a flow chart of step S150 in the hot water boiler automatic control method according to an embodiment of the present application. Figure 2 As shown, the step S150 includes: S151, time step segmentation of the time step data set of the water temperature data to obtain a sequence of local time step data sets of water temperature data; S152, multi-scale time series implicit encoding of the sequence of local time step data sets of water temperature data to obtain a sequence of multi-scale time series implicit correlation features of water temperature data; S153, dynamic aggregation of water temperature data time series features on the sequence of multi-scale time series implicit correlation features of water temperature data to obtain water temperature data time series dynamic aggregation features; S154, based on the water temperature data time series dynamic aggregation features, obtaining the proportional coefficient, differential coefficient and integral coefficient of the PID controller.

[0032] In step S151, the time step data set of the water temperature data is time-step segmented to obtain a sequence of local time step data sets of the water temperature data. Specifically, in an embodiment of the present application, the step S151 includes: the time step data set of the water temperature data is time-step segmented according to a predetermined time scale to obtain a sequence of local time step data sets of the water temperature data. It should be understood that the time step data set of the water temperature data exhibits different patterns and dynamic changes in different local time periods. Therefore, in order to be able to understand and analyze the temporal characteristic change information in each local time period more accurately and meticulously in the future, in the technical solution of the present application, the time step data set of the water temperature data is time-step segmented according to a predetermined time scale to divide the water temperature data within an entire time range into multiple local time segments, and obtain a sequence of local time step data sets of the water temperature data.

[0033] Time-step segmentation of the time-step data set of water temperature data can lay a solid foundation for the subsequent in-depth analysis of the water temperature variation law, accurate control of the burner firepower, and efficient water temperature regulation. The following is a detailed description of its implementation process: The core of time-step segmentation is to divide the entire time-step data set according to a predetermined time scale, thereby obtaining a series of local time-step data sets of water temperature data. The determination of this predetermined time scale requires comprehensive consideration of multiple key factors to ensure that the segmentation results can not only accurately reflect the local variation characteristics of water temperature, but also provide sufficient information granularity for subsequent analysis.

[0034] From the operating characteristics of hot water boilers, the change of water temperature is not completely random and disordered, but is affected by a variety of internal and external factors, showing a certain periodicity and trend. For example, in the actual operation of industrial hot water boilers, with the periodic changes in production tasks, the use of hot water will fluctuate regularly, which directly leads to the corresponding changes in the boiler water temperature. During the peak production period, a large amount of hot water is quickly taken, and the boiler needs to quickly replenish heat. The water temperature may show a more obvious downward trend in a short period of time; while during the production trough period, the use of hot water is reduced, and the water temperature may gradually rise. In addition, the heating and heat dissipation process of the boiler itself also has certain time characteristics. The working cycle of the heating element, the interaction between the heat dissipation rate and the ambient temperature, etc., will leave a unique mark in the change of water temperature.

[0035] Therefore, in practical applications, a multi-scale time step segmentation strategy is usually adopted according to specific needs and the operating characteristics of the hot water boiler. For example, a preliminary segmentation is first performed at a shorter time scale (such as 5 minutes) to obtain information on rapid changes in water temperature in a short period of time for real-time monitoring and immediate control adjustment; and then a secondary segmentation is performed at a longer time scale (such as 1 hour) to analyze the long-term trend of water temperature from a macro perspective, providing data support for the long-term performance optimization and strategic decision-making of the system.

[0036] Generally speaking, in the specific operation of time step segmentation, advanced algorithms and data processing techniques are needed to ensure the accuracy and efficiency of segmentation. A commonly used method is the sliding window algorithm, which uses a predetermined time scale as the window size and slides on the time step data set in sequence, sliding one time step at a time, thereby dividing the data set into a series of overlapping local time step data sets. The cleverness of this overlapping design is that it can not only preserve the continuity and correlation between adjacent time segments, avoiding the loss of important transition information due to abrupt segmentation, but also increase the sample size of the data, improving the reliability and stability of subsequent analysis.

[0037] For example, suppose there is a time step dataset that records the change of water temperature within 1 hour, and it is segmented by sliding window according to the predetermined time scale of 10 minutes. First, the data of the first 10 minutes are selected from the starting position of the dataset as the first local time step dataset; then, the window slides back 1 minute, and the data from the 2nd minute to the 11th minute are selected as the second local time step dataset, and so on, until the window slides to the end of the dataset. In this way, a series of local time step datasets with a length of 10 minutes and overlapping with each other are obtained, each of which contains detailed information about the change of water temperature during that time period.

[0038] By reasonably selecting the predetermined time scale and using an efficient segmentation algorithm, the time step data set of the water temperature data is segmented into time steps. The resulting sequence of local time step data sets of the water temperature data can provide multi-dimensional and multi-level water temperature change information for the hot water boiler automation control system, enabling the system to understand and grasp the dynamic characteristics of the boiler water temperature more comprehensively and deeply, thereby achieving more intelligent and precise water temperature control, ensuring that the hot water boiler always maintains efficient and stable operation under various complex working conditions.

[0039] In step S152, the sequence of the local time step data set of the water temperature data is subjected to multi-scale temporal implicit coding to obtain a sequence of multi-scale temporal implicit association features of the water temperature data. Specifically, in an embodiment of the present application, the step S152 includes: inputting each local time step data set of the water temperature data in the sequence of the local time step data set of the water temperature data into a sequence encoder based on the LSTM-RNN hybrid model to obtain a sequence of multi-scale temporal implicit association coding vectors of the water temperature data as a sequence of multi-scale temporal implicit association features of the water temperature data. It should be understood that each local time step data set of the water temperature data contains temporal feature information of different time scales, such as water temperature fluctuations in a short period of time and overall change patterns in a long period of time, and there are hidden correlations within the data. Based on this, in the technical solution of the present application, each local time step data set of the water temperature data in the sequence of the local time step data set of the water temperature data is input into a sequence encoder based on the LSTM-RNN hybrid model to obtain a sequence of multi-scale temporal implicit association coding vectors of the water temperature data. It is understandable that LSTM (Long Short-Term Memory Network), as a special type of RNN (Recurrent Neural Network), effectively solves the limitations of dealing with long-term dependency problems by introducing "memory units" and input gates, output gates, and forget gate mechanisms. In the water temperature control of hot water boilers, LSTM is able to capture the long-term trend of temperature changes over time and ensure the stable operation of the system. In contrast, standard RNNs are good at dealing with rapid changes or local patterns in the short term and can capture finer time series details. Therefore, in the LSTM-RNN hybrid model, the two complement each other: LSTM is responsible for learning long-term dependencies, while RNN supplements the analysis of short-term dynamics and local patterns. This combination not only better understands and models the temporal dynamic behavior of the hot water boiler system, but also improves the accuracy of predicting future water temperature changes, thereby providing strong support for the automated control of hot water boilers.

[0040] Inputting the sequence of local time step data sets of water temperature data into the sequence encoder based on the LSTM-RNN hybrid model to obtain the sequence of multi-scale temporal implicit association encoding vectors of water temperature data is one of the key links to achieve precise water temperature control. This process integrates advanced neural network technology, aiming to deeply explore the temporal characteristics and implicit associations in water temperature data, and provide rich and accurate information for subsequent control decisions. The following is a detailed description of its implementation process: As the core architecture of the sequence encoder, the LSTM-RNN hybrid model gives full play to the advantages of the two neural networks and provides strong support for the encoding of water temperature data. Recurrent neural network (RNN) has a unique memory function, which can process the temporal information in the sequence data and model the dependency between adjacent time steps. When processing water temperature data, RNN can capture the rapid changes and local fluctuation patterns of water temperature in the short term. For example, when the load of the hot water boiler suddenly increases or decreases, the water temperature rises or falls sharply in a short period of time. RNN can keenly perceive and record these changes, providing local dynamic information for subsequent analysis.

[0041] However, RNN has certain limitations when dealing with long-term dependencies, and is prone to gradient vanishing or explosion problems, resulting in inaccurate grasp of water temperature trends over long time spans. The emergence of long short-term memory networks (LSTMs) has effectively solved this problem. By introducing carefully designed memory units and gating mechanisms, LSTMs can selectively memorize or forget information, thereby better handling long-term dependencies. In the context of hot water boiler temperature control, LSTM can learn the changing trends of water temperature over a longer time range, such as the gradual change of boiler water temperature over hours or even days as the seasons change or production plans are adjusted. This accurate grasp of long-term trends is of vital importance for predicting future water temperature trends, adjusting burner firepower in advance, and optimizing overall system performance.

[0042] When the local time step data sets of each water temperature data are input into the sequence encoder based on the LSTM-RNN hybrid model, the data first enters the RNN layer. The neurons in the RNN layer process the data in sequence according to the time step. Each neuron receives the water temperature data of the current time step and the hidden state of the previous time step as input, and calculates the hidden state of the current time step according to the internal weight matrix and activation function. In this process, RNN gradually captures the short-term dynamic features in the water temperature data, such as rapid fluctuations and periodic changes in water temperature, by continuously updating the hidden state. These short-term features are encoded in the hidden state of the RNN layer, providing basic information for the subsequent processing of the LSTM layer.

[0043] After being processed by the RNN layer, the data is passed to the LSTM layer. The memory cells and gating mechanisms in the LSTM layer begin to work, deeply analyzing and processing the input data. The input gate is responsible for controlling the inflow of new information and determining which information will be written into the memory cell; the forget gate manages the degree of information forgetting in the memory cell, selectively retaining or discarding past information based on the current input and hidden state; the output gate determines which information to output as the hidden state of the current time step based on the current state and hidden state of the memory cell. Through this sophisticated information management mechanism, LSTM can avoid gradient problems caused by long-term dependencies while retaining the long-term water temperature change trend, thereby more accurately capturing the complex patterns and inherent connections in the water temperature data.

[0044] After the LSTM layer processes the data, the output layer of the model will generate a sequence of implicitly associated coding vectors of multi-scale time series of water temperature data according to specific task requirements. These coding vectors are an abstract representation of the original water temperature data in multi-scale time series. They not only contain the changing characteristics of water temperature at different time scales, but also contain implicit correlation information between data. For example, on a longer time scale, the coding vector may reflect the trend of water temperature changing with the seasons; on a shorter time scale, it may reveal the relationship between the rapid fluctuation of water temperature within a few minutes and the instantaneous fire adjustment of the burner.

[0045] In order to ensure that the sequence encoder can accurately learn the features and associations in the water temperature data, the model needs to be trained extensively. The training process usually adopts a supervised learning method, using historical water temperature data and corresponding burner control data as training samples. By minimizing the error function between the predicted results and the actual control data, the internal weight matrix and bias parameters are continuously adjusted to improve the encoding accuracy and prediction ability of the water temperature data. During the training process, some advanced optimization algorithms, such as stochastic gradient descent (SGD) and its variants Adagrad, Adadelta, Adam, etc., can also be used to accelerate the convergence speed of the model and improve training efficiency.

[0046] By inputting the sequence of local time step data sets of water temperature data into the trained sequence encoder based on the LSTM-RNN hybrid model, the obtained sequence of multi-scale temporal implicit association encoding vectors of water temperature data can provide high-quality feature input data for subsequent data processing operations.

[0047] In step S153, the sequence of the multi-scale time series implicit correlation features of the water temperature data is dynamically aggregated to obtain the dynamic aggregation features of the water temperature data time series. Specifically, Figure 3 FIG. 1 is a flow chart of step S153 in the hot water boiler automatic control method according to an embodiment of the present application. Figure 3As shown, the step S153 includes: S1531, performing characteristic static potential energy measurement on each multi-scale time series implicitly associated coding vector of water temperature data in the sequence of multi-scale time series implicitly associated coding vectors of water temperature data to obtain a sequence of multi-scale time series characteristic static potential energy measurement coefficients of water temperature data; S1532, based on the sequence of multi-scale time series characteristic static potential energy measurement coefficients of water temperature data, performing weighted aggregation on the sequence of multi-scale time series implicitly associated coding vectors of water temperature data to obtain a multi-scale time series pseudo-anchor aggregation center representation vector of water temperature data; S1533, based on the multi-scale time series pseudo-anchor aggregation center representation vector of water temperature data, performing dynamic aggregation on the sequence of multi-scale time series implicitly associated coding vectors of water temperature data along the direction of motion to obtain a water temperature data time series dynamic aggregation coding vector as the water temperature data time series dynamic aggregation feature. In particular, in an embodiment of the present application, the static potential energy measurement coefficient of the multi-scale time series characteristics of water temperature data of the multi-scale time series implicit association coding vectors of each water temperature data is related to the mean and standard deviation of the multi-scale time series implicit association coding vectors of each water temperature data.

[0048] It should be understood that the implicit correlation features of the multi-scale time series of each water temperature data have different significance and force on the overall time axis, and at the same time, the degree of connection between each feature and the core essence is also different. Therefore, in order to integrate the significant feature information on different time scales, to better understand the complex dynamic behavior and internal patterns of water temperature changes, to generate a compact and expressive representation that can fully reflect the dynamic behavior of the hot water boiler system, to intelligently adjust the proportional, integral and differential coefficients of the PID controller, and to achieve the optimal burner firepower regulation, in the technical solution of the present application, the sequence of the multi-scale time series implicit correlation features of the water temperature data is dynamically aggregated to obtain the dynamic aggregation features of the water temperature data time series. In this way, it can ensure that the importance of each feature is properly reflected. At the same time, attention is paid to the interaction and causal relationship between features to reveal deeper system dynamics.

[0049] Specifically, each water temperature data multi-scale time series implicit association coding vector in the sequence of water temperature data multi-scale time series implicit association coding vectors is first subjected to characteristic static potential energy measurement to obtain a sequence of water temperature data multi-scale time series characteristic static potential energy measurement coefficients. The above process can be expressed as: ; in, is a sequence of multi-scale time series implicit association coding vectors of the water temperature data, and are the first, second, and third vectors in the sequence of multi-scale time series implicit association coding vectors of the water temperature data. and Multi-scale time series implicit correlation coding vector of water temperature data, It is The eigenvalues ​​of each position in the multi-scale time series implicit association encoding vector of water temperature data, It is The mean of the multi-scale time series implicit correlation coding vector of water temperature data, It is The standard deviation of the multi-scale time series implicit correlation encoding vector of water temperature data, It is The number of eigenvalues ​​in the multi-scale time series implicit correlation coding vector of water temperature data, It is the first in the sequence of the static potential energy measurement coefficients of the multi-scale time series characteristics of water temperature data. The static potential energy measurement coefficient of the multi-scale time series characteristics of water temperature data.

[0050] It should be understood that by measuring the characteristic static potential energy of each water temperature data multi-scale time series implicit association coding vector in the sequence of the water temperature data multi-scale time series implicit association coding vector, the potential energy level of each water temperature data multi-scale time series implicit association coding vector under static conditions can be evaluated, so that the intrinsic connection between different water temperature data time series vectors and the correlation with the target output can be better captured, and then each water temperature data multi-scale time series implicit association coding vector can be assigned a measurement coefficient of importance and influence under the current water temperature data distribution. The sequence of static potential energy measurement coefficients of multi-scale time series characteristics of water temperature data generated actually reflects the trend strength of the water temperature time series characteristics to maintain the status quo or change without external interference, which can provide important basic information for subsequent processing.

[0051] Specifically, in the embodiment of the present application, the step S1532 includes: performing feature energy level gating screening on the sequence of implicit association coding vectors of the multi-scale time series of water temperature data to obtain a sequence of static potential energy weight factors of the multi-scale time series features of water temperature data. The process can be expressed as: ; in, It is the first in the sequence of the static potential energy measurement coefficients of the multi-scale time series characteristics of the standardized water temperature data. The static potential energy measurement coefficient of the multi-scale time series characteristics of the standardized water temperature data, is the preset threshold, is a masking operation, It is the first in the sequence of static potential energy weight factors of multi-scale time series characteristics of water temperature data. The static potential energy weight factor of the multi-scale time series characteristics of water temperature data.

[0052] The position-weighted sum of the sequence of static potential energy weight factors of the multi-scale time series characteristics of the water temperature data and the sequence of implicit association coding vectors of the multi-scale time series of the water temperature data is calculated to obtain the pseudo-anchored aggregation center representation vector of the multi-scale time series of the water temperature data. The process can be expressed as: ;in, is the sequence of the multi-scale time series implicit association coding vector of the water temperature data. Multi-scale time series implicit correlation coding vector of water temperature data, It is the first in the sequence of static potential energy weight factors of multi-scale time series characteristics of water temperature data. The static potential energy weight factor of the multi-scale time series characteristics of water temperature data, is the number of vectors in the sequence of multi-scale time series implicit association coding vectors of water temperature data, It is the pseudo-anchor aggregation center representation vector of multi-scale time series of water temperature data.

[0053] It should be understood that when processing the sequence of multi-scale temporal implicitly associated coding vectors of water temperature data, it often contains a large amount of information of different dimensions, and not all temporal implicitly associated coding vectors have the same important role in the final analysis result of the PID coefficient. By performing feature energy level gating screening on the sequence of multi-scale temporal implicitly associated coding vectors of water temperature data, those vectors that really contribute to the final result can be accurately found, so that subsequent analysis, processing and other operations can be carried out around these key features, avoiding irrelevant or unimportant information from interfering with the entire processing process, which helps to make the entire processing process more efficient and the results more reliable. Next, it is necessary to calculate the weighted sum of the sequence of the static potential energy weight factors of the multi-scale temporal characteristics of the water temperature data and the sequence of the multi-scale temporal implicitly associated coding vectors of the water temperature data to obtain the multi-scale temporal pseudo-anchored aggregation center representation vector of the water temperature data. That is, a center point representing the overall trend of the water temperature data can be generated through the weighted summation operation, that is, the multi-scale temporal pseudo-anchored aggregation center representation vector of the water temperature data. The generated center representation vector can guide the subsequent feature vector aggregation process. In this way, we can clearly see how each water temperature data multi-scale temporal implicit correlation encoding vector is affected by the center point, and predict the movement trajectory of the water temperature data multi-scale temporal implicit correlation encoding vector in the feature space.

[0054] Specifically, in the embodiment of the present application, the step S1533 includes: calculating the aggregation movement direction of each water temperature data multi-scale time series implicit association coding vector in the sequence of the water temperature data multi-scale time series implicit association coding vector relative to the water temperature data multi-scale time series pseudo-anchor aggregation center representation vector to obtain a sequence of water temperature data aggregation running directions. This process can be expressed as: ;in, is the sequence of the multi-scale time series implicit association coding vector of the water temperature data. Multi-scale time series implicit correlation coding vector of water temperature data, is the pseudo-anchor aggregation center representation vector of multi-scale time series of water temperature data, is the length of the vector, is the inverse cosine function, It is the first in the sequence of the water temperature data aggregation operation direction. The water temperature data aggregation operation direction.

[0055] Based on the sequence of the water temperature data aggregation running direction, the sequence of the water temperature data multi-scale time series implicit association coding vectors is dynamically aggregated toward the water temperature data multi-scale time series pseudo-anchor aggregation center representation vector to obtain the water temperature data time series dynamic aggregation coding vector. This process can be expressed as: ;in, is the sequence of the multi-scale time series implicit association coding vector of the water temperature data. Multi-scale time series implicit correlation coding vector of water temperature data, is the number of vectors in the sequence of multi-scale time series implicit association coding vectors of water temperature data, is the pseudo-anchor aggregation center representation vector of multi-scale time series of water temperature data, It is the first in the sequence of the water temperature data aggregation operation direction. The water temperature data aggregation operation direction, is a sine function, is the first in the sequence of weight matrices A weight matrix, is the first in the sequence of bias vectors A bias vector, It is the time series dynamic aggregation coding vector of the water temperature data.

[0056] It should be understood that by calculating the aggregation movement direction of each water temperature data multi-scale time series implicit association coding vector in the sequence of the water temperature data multi-scale time series implicit association coding vector relative to the water temperature data multi-scale time series pseudo-anchor aggregation center representation vector, the change trend of each water temperature data multi-scale time series implicit association coding vector relative to the center point can be quantified, which helps to more accurately capture the dynamic evolution characteristics of water temperature data under multi-scale time series. At the same time, by determining the aggregation movement direction of these vectors relative to the aggregation center representation vector, the dynamic relationship within the feature space in the feature space where the water temperature data is located can be clearly revealed. For example, it can be understood whether the multi-scale time series implicit association coding vectors of different temperature data are gradually approaching or moving away from the center point, and then the interaction and coordinated changes between them can be known, which helps to deeply explore the hidden inherent laws of water temperature data at different time series scales. In addition, understanding the aggregation movement direction of these vectors is also helpful to explore the evolution path of water temperature data in the feature space, which helps to grasp the overall performance of water temperature data characteristics from a macro level. Finally, based on the sequence of the water temperature data aggregation running direction, the sequence of the water temperature data multi-scale time series implicit association coding vectors is dynamically aggregated toward the water temperature data multi-scale time series pseudo-anchor aggregation center representation vector to obtain the water temperature data time series dynamic aggregation coding vector. That is, according to the previously calculated sequence of the water temperature data aggregation running direction and the static potential energy measurement coefficients of each water temperature data multi-scale time series feature, the initial water temperature data multi-scale time series implicit association coding vector is adjusted and optimized based on the simulated dynamic model, so that it is more closely around the water temperature data multi-scale time series pseudo-anchor aggregation center representation vector to form a new water temperature data time series dynamic aggregation coding vector. In this process, the interaction and adjustment between different vectors help to discover the deep patterns hidden in the water temperature data, which helps to reveal hidden feature relationships and enhance the feature expressiveness of the original data.

[0057] In step S154, based on the dynamic aggregation characteristics of the water temperature data time series, the proportional coefficient, differential coefficient and integral coefficient of the PID controller are obtained. Specifically, in an embodiment of the present application, the step S154 includes: inputting the water temperature data time series dynamic aggregation coding vector into a decoder-based PID controller parameter optimization module to obtain the proportional coefficient, differential coefficient and integral coefficient of the PID controller. It should be understood that the water temperature data time series dynamic aggregation coding vector contains rich time series information mined from the water temperature data time step data set, which reflects the dynamic characteristics of water temperature over time, including the fluctuation trend and periodic law of water temperature. In order to make full use of these time series characteristics to dynamically adjust the parameters of the PID controller, it is necessary to input the water temperature data time series dynamic aggregation coding vector into the decoder-based PID controller parameter optimization module. Specifically, the decoder is a neural network model that can decode and output parameters that can be directly used for the PID controller based on the relationship between the hidden patterns and features in the learned water temperature data time series dynamic aggregation coding vector through a large amount of data training. The decoded proportional coefficient of the PID controller determines the direct proportional relationship between the control output (burner power adjustment value) and the real-time deviation value of the water temperature. The appropriate proportional coefficient can find a balance between control accuracy (fast response to water temperature deviation) and system stability. The differential coefficient can adjust the control output according to the rate of change of the water temperature. For example, if the water temperature is rising rapidly, the differential coefficient will appropriately reduce the calculated burner power adjustment value to avoid excessive rise in water temperature and exceed the target water temperature. This predictive control can effectively prevent drastic fluctuations in water temperature and improve the dynamic performance of the system. The integral coefficient is mainly used to eliminate the steady-state error of the system. During the operation of the hot water boiler, even if the proportional control and differential control can make the water temperature close to the target water temperature, there may be a small deviation due to various factors (such as incomplete combustion of the burner, heat loss, etc.). The integral control will adjust the control output according to the accumulation of this deviation. For example, if the water temperature has been slightly lower than the target water temperature, the integral coefficient will gradually increase the burner power until the steady-state error is eliminated and the water temperature stabilizes at the target water temperature. It takes into account the cumulative effect of water temperature deviation over time and can adjust the control strategy based on historical water temperature deviation information. This is very effective for dealing with long-standing, small water temperature deviations, and helps to improve the long-term stability and accuracy of water temperature control. In this way, by adjusting the P, I, and D parameters of the PID controller in real time, it can ensure that it is always in the optimal state and maintains good control performance even when working conditions change. It can also capture the nonlinear behavior and complex patterns of the system, allowing the controller to automatically adapt to new dynamic situations, avoiding the problem that traditional fixed parameters cannot cope with changes.

[0058] In a preferred example, considering that each water temperature data multi-scale time series implicit association coding vector in the sequence of the water temperature data multi-scale time series implicit association coding vectors respectively represents the short-range and long-range bidirectional time series correlation characteristics of the water temperature data in the local time domain, when performing feature aggregation based on feature dynamics analysis, the insufficient dynamic correspondence of the time series features in different local time domains will lead to insufficient representation of the long-distance aggregation features of the water temperature data time series dynamic aggregation coding vector, thereby affecting the accuracy of the proportional coefficient, differential coefficient and integral coefficient of the PID controller obtained by inputting the decoder-based PID controller parameter optimization module.

[0059] Based on this, before inputting the water temperature data time series dynamic aggregation coding vector into the decoder-based PID controller parameter optimization module, the present application first optimizes the water temperature data time series dynamic aggregation coding vector, including the steps of arranging the various eigenvalues ​​of the water temperature data time series dynamic aggregation coding vector in order from small to large to form a water temperature data time series dynamic aggregation sequential coding vector.

[0060] In response to the water temperature data time series dynamic aggregation sequence encoding vector The eigenvalue and The absolute value of the difference between the eigenvalues ​​is less than or equal to the distance difference hyperparameter , calculate the The eigenvalues ​​are similar to the The weighted sum between the eigenvalues ​​is the optimized The eigenvalues ​​are: ;in, are the first and second order encoding vectors of the water temperature data time series dynamic aggregation order. The eigenvalue and Eigenvalues, For the optimized eigenvalues, is the first weight hyperparameter, is the second weight hyperparameter.

[0061] In response to the water temperature data time series dynamic aggregation sequence encoding vector The eigenvalue and The absolute value of the difference between the eigenvalues ​​is greater than the distance difference hyperparameter , calculate the square root of the sum of squares of all eigenvalues ​​of the water temperature data time series dynamic aggregation coding vector , multiplied by 2 and then divided by the length of the water temperature data time series dynamic aggregation coding vector The square of the water temperature data time series dynamic aggregation space primitive value , and multiply the water temperature data time series dynamic aggregation space primitive value by the first After the eigenvalue, calculate the product with the first The weighted reduction between the eigenvalues ​​is the optimized The eigenvalues ​​are: ;in, is the third weight hyperparameter, is the fourth weight hyperparameter.

[0062] The optimized first step of combining the water temperature data time series dynamic aggregation order encoding vector The eigenvalue is used to obtain an optimized water temperature data time series dynamic aggregation coding vector, wherein the first eigenvalue of the water temperature data time series dynamic aggregation sequence coding vector remains unchanged.

[0063] Accordingly, in order to address the problem of insufficient global interactive response representation capability of the feature set of the water temperature data time series dynamic aggregation coding vector under a predetermined eigenvalue sequential distribution due to a long distance exceeding a predetermined local distribution interval threshold, a high-dimensional feature space primitive representation based on self-inner product fusion of the water temperature data time series dynamic aggregation coding vector is used to capture the complex structure of the global network interaction of its eigenvalues, thereby reconstructing the interactive response relationship between the eigenvalues ​​of the water temperature data time series dynamic aggregation coding vector by simulating scale-based high-dimensional feature space potential primitives, so as to achieve coding reconstruction of the real sequence distribution behavior of the water temperature data time series dynamic aggregation coding vector under long distance, improve the interactive response expression effect of the water temperature data time series dynamic aggregation coding vector, and improve the accuracy of the proportional coefficient, differential coefficient and integral coefficient of the PID controller obtained by inputting the water temperature data time series dynamic aggregation coding vector into the decoder-based PID controller parameter optimization module.

[0064] In summary, the proportional coefficient, differential coefficient and integral coefficient of the PID controller are determined based on the time step data set of the water temperature data and are clearly explained. It uses artificial intelligence-based data processing and analysis technology to perform time step segmentation and multi-scale time series association coding on the water temperature data, so as to intelligently generate the proportional coefficient, differential coefficient and integral coefficient of the PID controller according to the multi-scale time series implicit coding of the sequence of the local time step data set of the water temperature data to obtain the sequence characteristics of the multi-scale time series implicit association characteristics of the water temperature data. In this way, the P, I, and D parameters of the PID controller can be adjusted in real time to ensure that it is always in the optimal state and maintain good control performance even when the working conditions change. And it can capture the nonlinear behavior and complex patterns of the system, so that the controller can automatically adapt to new dynamic situations, avoiding the problem that traditional fixed parameters cannot cope with changes.

[0065] In step S160, the burner is controlled based on the control instruction. It should be understood that the burner is a key device for heating water in a hot water boiler. By controlling the firepower of the burner, the heat absorbed by the hot water can be changed, thereby adjusting the water temperature. That is, by adjusting the burner firepower adjustment value, the water temperature in the hot water boiler can be changed in a desired manner, that is, approaching the target water temperature set by the user and finally stabilizing near the target water temperature.

[0066] In summary, the automatic control method of the hot water boiler based on the embodiment of the present application is explained, which first collects water temperature data in real time to construct a time step data set of water temperature data, and extracts the current water temperature information therefrom, and simultaneously obtains the target water temperature data set set by the user from the background database, and then calculates the real-time deviation value between the current water temperature and the target water temperature, and inputs this deviation value into the PID controller to generate a control instruction including the burner firepower adjustment value, and the proportional coefficient, differential coefficient and integral coefficient of the controller are dynamically determined according to the time step data set of the water temperature data, and finally, the burner is precisely controlled according to this control instruction to realize the real-time adjustment of the water temperature. In this way, the automatic and intelligent adjustment and control of the water temperature of the hot water boiler can be realized.

[0067] In an embodiment of the present application, a hot water boiler automatic control system is also provided, which is used to execute the hot water boiler automatic control method as described above. Figure 4 : is a system block diagram of a hot water boiler automatic control system according to an embodiment of the present application. Figure 4 As shown, according to the hot water boiler automatic control system 100 of the embodiment of the present application, it includes: a water temperature time series data segmentation module 110, which is used to perform time step segmentation on the time step data set of the water temperature data to obtain a sequence of local time step data sets of water temperature data; a water temperature time series data encoding module 120, which is used to perform multi-scale time series implicit encoding on the sequence of local time step data sets of water temperature data to obtain a sequence of multi-scale time series implicit correlation features of water temperature data; a water temperature time series feature dynamic aggregation module 130, which is used to perform water temperature data time series feature dynamic aggregation on the sequence of multi-scale time series implicit correlation features of water temperature data to obtain water temperature data time series dynamic aggregation features; a PID controller coefficient generation module 140, which is used to obtain the proportional coefficient, differential coefficient and integral coefficient of the PID controller based on the water temperature data time series dynamic aggregation features.

[0068] Here, those skilled in the art will appreciate that the specific functions and operations of the various units and modules in the above hot water boiler automation control system 100 have been described in detail above. Figures 1 to 3 The hot water boiler automatic control method has been described in detail, and therefore, its repeated description will be omitted.

[0069] In summary, the hot water boiler automatic control system 100 based on the embodiment of the present application is explained, which first collects water temperature data in real time to construct a time step data set of water temperature data, and extracts the current water temperature information therefrom, and simultaneously obtains the target water temperature data set set by the user from the background database, and then calculates the real-time deviation value between the current water temperature and the target water temperature, and inputs this deviation value into the PID controller to generate a control instruction including the burner firepower adjustment value, and the proportional coefficient, differential coefficient and integral coefficient of the controller are dynamically determined according to the time step data set of the water temperature data, and finally, the burner is precisely controlled according to this control instruction to realize the real-time adjustment of the water temperature. In this way, the automatic and intelligent adjustment and control of the water temperature of the hot water boiler can be realized.

Claims

1. A hot water boiler automatic control method, characterized in that: include: Collect water temperature data in real time to obtain a time step data set of water temperature data; Extracting current water temperature data from the time step data set of the water temperature data; Extract target water temperature data set by the user from the background database; Calculating a real-time water temperature deviation value between the target water temperature data and the current water temperature data; The real-time deviation value of water temperature is input into a PID controller to obtain a control instruction, and the control instruction includes a burner firepower adjustment value, wherein the proportional coefficient, differential coefficient and integral coefficient of the PID controller are determined based on the time step data set of the water temperature data; the burner is controlled based on the control instruction; wherein the proportional coefficient, differential coefficient and integral coefficient of the PID controller are determined based on the time step data set of the water temperature data, including: time step segmentation of the time step data set of the water temperature data to obtain a sequence of local time step data sets of water temperature data; multi-scale time series implicit encoding of the sequence of local time step data sets of water temperature data to obtain a sequence of multi-scale time series implicit correlation features of water temperature data; dynamic aggregation of water temperature data time series features of the sequence of multi-scale time series implicit correlation features of water temperature data to obtain water temperature data time series dynamic aggregation features; based on the water temperature data time series dynamic aggregation features, the proportional coefficient of the PID controller is obtained. Coefficients, differential coefficients and integral coefficients; wherein, the sequence of the multi-scale time series implicit correlation features of the water temperature data is dynamically aggregated to obtain the dynamic aggregation features of the water temperature data time series, including: performing characteristic static potential energy measurement on each water temperature data multi-scale time series implicit correlation coding vector in the sequence of the multi-scale time series implicit correlation coding vector of the water temperature data to obtain a sequence of static potential energy measurement coefficients of the multi-scale time series features of the water temperature data; based on the sequence of static potential energy measurement coefficients of the multi-scale time series features of the water temperature data, weighted aggregation is performed on the sequence of the multi-scale time series implicit correlation coding vector of the water temperature data to obtain a multi-scale time series pseudo-anchor aggregation center representation vector of the water temperature data; based on the multi-scale time series pseudo-anchor aggregation center representation vector of the water temperature data, dynamic aggregation is performed on the sequence of the multi-scale time series implicit correlation coding vector of the water temperature data along the direction of motion to obtain a water temperature data time series dynamic aggregation coding vector as the water temperature data time series dynamic aggregation feature.

2. The hot water boiler automatic control method according to claim 1, characterized in that: The time step data set of the water temperature data is time-step segmented to obtain a sequence of local time step data sets of the water temperature data, including: the time step data set of the water temperature data is time-step segmented according to a predetermined time scale to obtain a sequence of local time step data sets of the water temperature data.

3. The hot water boiler automatic control method according to claim 2, characterized in that: The sequence of the local time step data sets of water temperature data is subjected to multi-scale time series implicit encoding to obtain a sequence of multi-scale time series implicit correlation features of water temperature data, including: inputting each local time step data set of water temperature data in the sequence of the local time step data sets of water temperature data into a sequence encoder based on the LSTM-RNN hybrid model to obtain a sequence of multi-scale time series implicit correlation encoding vectors of water temperature data as the sequence of multi-scale time series implicit correlation features of water temperature data.

4. The hot water boiler automatic control method according to claim 3, characterized in that: The water temperature data multi-scale time series characteristic static potential energy measurement coefficient of the multi-scale time series implicit association coding vector of each water temperature data is related to the mean and standard deviation of the multi-scale time series implicit association coding vector of each water temperature data.

5. The hot water boiler automatic control method according to claim 4, characterized in that: Based on the sequence of static potential energy measurement coefficients of the multi-scale time series characteristics of the water temperature data, the sequence of the multi-scale time series implicit association coding vectors of the water temperature data is weightedly aggregated to obtain the multi-scale time series pseudo-anchored aggregation center representation vector of the water temperature data, including: performing feature energy level gating screening on the sequence of the multi-scale time series implicit association coding vectors of the water temperature data to obtain a sequence of static potential energy weight factors of the multi-scale time series characteristics of the water temperature data; calculating the position-weighted sum of the sequence of static potential energy weight factors of the multi-scale time series characteristics of the water temperature data and the sequence of the multi-scale time series implicit association coding vectors of the water temperature data to obtain the multi-scale time series pseudo-anchored aggregation center representation vector of the water temperature data.

6. The hot water boiler automatic control method according to claim 5, characterized in that: Based on the multi-scale time series pseudo-anchored aggregation center representation vector of the water temperature data, the sequence of the multi-scale time series implicitly associated coding vectors of the water temperature data is dynamically aggregated along the movement direction to obtain the water temperature data time series dynamic aggregation coding vector, including: calculating the aggregation movement direction of each multi-scale time series implicitly associated coding vector of the water temperature data in the sequence of the multi-scale time series implicitly associated coding vector of the water temperature data relative to the multi-scale time series pseudo-anchored aggregation center representation vector of the water temperature data to obtain a sequence of aggregated running directions of the water temperature data; based on the sequence of aggregated running directions of the water temperature data, the sequence of the multi-scale time series implicitly associated coding vector of the water temperature data is dynamically aggregated toward the multi-scale time series pseudo-anchored aggregation center representation vector of the water temperature data to obtain the water temperature data time series dynamic aggregation coding vector.

7. The hot water boiler automatic control method according to claim 6, characterized in that: Based on the time series dynamic aggregation characteristics of the water temperature data, the proportional coefficient, differential coefficient and integral coefficient of the PID controller are obtained, including: inputting the time series dynamic aggregation coding vector of the water temperature data into a decoder-based PID controller parameter optimization module to obtain the proportional coefficient, differential coefficient and integral coefficient of the PID controller.

8. A hot water boiler automatic control system, used to execute the hot water boiler automatic control method according to any one of claims 1 to 7, characterized in that: The automatic control system of a hot water boiler comprises: a water temperature time series data segmentation module, which is used to perform time step segmentation on the time step data set of the water temperature data to obtain a sequence of local time step data sets of the water temperature data; a water temperature time series data encoding module, which is used to perform multi-scale time series implicit encoding on the sequence of local time step data sets of the water temperature data to obtain a sequence of multi-scale time series implicit correlation features of the water temperature data; a water temperature time series feature dynamic aggregation module, which is used to perform water temperature data time series feature dynamic aggregation on the sequence of multi-scale time series implicit correlation features of the water temperature data to obtain water temperature data time series dynamic aggregation features; a PID controller coefficient generation module, which is used to obtain the proportional coefficient, differential coefficient and integral coefficient of the PID controller based on the water temperature data time series dynamic aggregation features.

9. The hot water boiler automatic control system according to claim 8, characterized in that: The water temperature time series data segmentation module is used to: segment the time step data set of the water temperature data according to a predetermined time scale to obtain a sequence of local time step data sets of the water temperature data.

Images