An electronic incense burner adaptive PID temperature control method based on artificial intelligence

Abstract

This invention relates to the field of temperature control technology, specifically to an adaptive PID temperature control method for electronic incense burners based on artificial intelligence. The method includes: acquiring the global convexity coefficient, global thermal inertia coefficient, convexity coefficient variance, and thermal inertia coefficient variance at the time to be analyzed and monitored; adjusting the baseline P, I, and D parameters based on the global convexity coefficient and global thermal inertia coefficient to obtain the P, I, and D parameters to be determined at the t-th time to be analyzed and monitored; determining whether the target inflection point and characteristic convergence point have occurred at the t-th time to be analyzed and monitored based on the original furnace temperature in the neighborhood of the t-th time to be analyzed and monitored, as well as the convexity coefficient variance and thermal inertia coefficient variance; and adaptively generating the controller output signal at the t-th time to be analyzed and monitored based on the determination result and the P, I, and D parameters to be determined, and driving the execution. This invention can improve the temperature control effect.

Description

Technical Field

[0001] This invention relates to the field of temperature control technology, and specifically to an adaptive PID temperature control method for electronic incense burners based on artificial intelligence. Background Technology

[0002] As a modern incense burning device, the core working principle of an electronic incense burner is to precisely heat the heating medium inside the burner using an electric heating element, causing the incense materials to release their aroma at a specific temperature, thus achieving a safe incense burning experience without open flame or smoke. Compared with traditional incense burning methods, electronic incense burners have significant advantages such as controllable temperature, high safety, and pure aroma, and have been widely used in quiet spaces such as homes, tea rooms, yoga studios, and meditation venues, becoming an important carrier for improving the quality of modern life.

[0003] Different fragrance materials have vastly different optimal evaporation temperature windows. For example, sandalwood essential oil is most mellow when evaporating between 120℃ and 150℃, while agarwood needs to be controlled between 160℃ and 200℃ to release its unique woody aroma. Some floral essential oils perform best in the low-temperature range of 80℃ to 100℃. If the actual temperature is lower than the optimal evaporation window, the fragrance material cannot fully release its aroma, resulting in resource waste and a decline in the user experience. If the temperature is too high, it may cause the fragrance material to burn and produce an unpleasant odor, or even cause fire hazards. Therefore, in order to ensure user experience and functionality, it is usually necessary to control the temperature of electronic incense burners.

[0004] In existing technologies, traditional proportional-integral-derivative (PID) control algorithms are commonly used to control the temperature of electronic incense burners. However, different incense materials have different thermodynamic properties, and traditional PID controllers cannot detect these differences. This leads to unstable or poor control effects under different incense materials and dosages. Therefore, how to improve the adaptive PID temperature control of electronic incense burners to enhance the temperature control effect has become an urgent problem to be solved. Summary of the Invention

[0005] To address the aforementioned problems, this invention provides an adaptive PID temperature control method for electronic incense burners based on artificial intelligence. The specific technical solution adopted is as follows:

[0006] One embodiment of the present invention provides an adaptive PID temperature control method for an electronic incense burner based on artificial intelligence, comprising the following steps:

[0007] Obtain the normalized furnace temperature corresponding to each normalized monitoring time within the normalized time window of the electronic incense burner at the time to be analyzed and monitored;

[0008] Based on the normalized monitoring time and normalized furnace temperature, the local convexity coefficient and local thermal inertia coefficient at the monitoring time to be analyzed are obtained. The local convexity coefficient and local thermal inertia coefficient are then estimated asymptotically to obtain the global convexity coefficient and global thermal inertia coefficient at different monitoring times to be analyzed. The variances of the local convexity coefficient and local thermal inertia coefficient at the first t monitoring times to be analyzed are denoted as the variances of the convexity coefficient and thermal inertia coefficient at the t-th monitoring time to be analyzed, respectively. Based on the global convexity coefficient and global thermal inertia coefficient, the baseline P, I, and D parameters are adjusted to obtain the P, I, and D parameters to be determined at the t-th monitoring time to be analyzed.

[0009] Based on the original furnace temperature in the neighborhood of the t-th monitoring time and the variance of the convexity coefficient and the variance of the thermal inertia coefficient, it is determined whether the target inflection point and characteristic convergence time occur at the t-th monitoring time. Based on the determination result and the P, I and D parameters to be determined, the controller output signal at the t-th monitoring time is adaptively generated and driven to execute.

[0010] Beneficial effects: This invention first obtains the normalized furnace temperature corresponding to each normalized monitoring time within the normalized time window of the electronic incense burner under the monitoring time to be analyzed; then, based on the normalized monitoring time and normalized furnace temperature, it obtains the local convexity coefficient and local thermal inertia coefficient under the monitoring time to be analyzed, and performs asymptotic recursive global estimation on the local convexity coefficient and local thermal inertia coefficient to obtain the global convexity coefficient and global thermal inertia coefficient under different monitoring times to be analyzed. The variance of the local convexity coefficient and the variance of the local thermal inertia coefficient under the first t monitoring times to be analyzed are respectively denoted as the t-th monitoring time to be analyzed. The variances of the convexity coefficient and thermal inertia coefficient at time t are used to adjust the baseline P, I, and D parameters based on the global convexity coefficient and global thermal inertia coefficient, resulting in the P, I, and D parameters to be determined at the t-th monitoring time. Finally, based on the original furnace temperature in the neighborhood of the t-th monitoring time and the variances of the convexity coefficient and thermal inertia coefficient, it is determined whether the target inflection point and feature convergence point have occurred at the t-th monitoring time. Based on the determination results and the P, I, and D parameters to be determined, the controller output signal at the t-th monitoring time is adaptively generated and driven for execution. Furthermore, this invention, through the feature extraction of convexity coefficient and thermal inertia coefficient, and the integration of the feature extraction results, the target inflection point, and feature convergence point determination results into an adaptive PID temperature control strategy, can eliminate steady-state errors, reduce overshoot and oscillations, and improve temperature control accuracy. In other words, this invention's adaptive PID temperature control strategy, based on feature extraction results and combined with the target inflection point and feature convergence point determination results, can improve temperature control performance. Attached Figure Description

[0011] To more clearly illustrate the technical solutions and advantages in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0012] Figure 1 This is a flowchart of an adaptive PID temperature control method for an electronic incense burner based on artificial intelligence, according to the present invention. Detailed Implementation

[0013] The technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. All other embodiments obtained by those skilled in the art based on the embodiments of the present invention are within the protection scope of the embodiments of the present invention.

[0014] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as commonly understood by one of ordinary skill in the art.

[0015] This embodiment provides an adaptive PID temperature control method for electronic incense burners based on artificial intelligence, detailed as follows:

[0016] like Figure 1 As shown, the adaptive PID temperature control method for the electronic incense burner includes the following steps:

[0017] Step S001: Obtain the normalized furnace temperature corresponding to each normalized monitoring time within the normalized time window of the electronic incense burner at the time to be analyzed and monitored.

[0018] In order to ensure the temperature control effect under different incense materials and different amounts, this embodiment will perform adaptive PID temperature control on the electronic incense burner by extracting features during the heating stage and integrating the feature extraction results, the target inflection point time, and the feature convergence time discrimination results. For ease of understanding, this embodiment will use the PID temperature control process in the current operation or current working cycle of the electronic incense burner as an example for description. The electronic incense burner analyzed in this embodiment is currently running, and all subsequent monitoring or running moments in this embodiment belong to the current operation or current working cycle. The current operation or current working cycle of the electronic incense burner refers to the process from the most recent power-on start-up to automatic or manual power-off. That is, a complete operation or a complete single working cycle of the electronic incense burner is the process from power-on start-up to automatic or manual power-off. The operation of the electronic incense burner is the process of releasing the type of incense material or aroma components through heating. Before each start-up, the electronic incense burner sets a target temperature. That is, the current operation of the electronic incense burner corresponds to a target temperature, which is the constant temperature control setting value of the heating plate surface of the electronic incense burner. In specific applications, implementers generally set the target temperature based on the physicochemical properties of the incense material, the volatilization kinetics, and the intelligent adaptation mechanism of the equipment. For example, if the incense material currently used is agarwood powder, then the target temperature can be set to 170 degrees Celsius.

[0019] This embodiment first uses sensors to collect real-time combustion chamber temperature and ambient temperature data during the operation of the electronic incense burner. The combustion chamber temperature collected by the sensors at each operating moment is recorded as the original combustion chamber temperature of the electronic incense burner at that corresponding operating moment, and the ambient temperature collected by the sensors at each operating moment is recorded as the ambient temperature of the electronic incense burner at that corresponding operating moment. That is, the original combustion chamber temperature is the combustion chamber temperature directly collected by the sensors. Then, all operating moments after the most recent power-on start-up moment are recorded as monitoring moments. Next, the monitoring moments after the most recent power-on start-up moment are iterated through, and the monitoring moment where the first temperature difference exceeds a preset temperature threshold is recorded as the initial temperature rise monitoring moment. Any monitoring moment... The temperature difference at the corresponding monitoring time is the difference between the original furnace temperature at the current time and the ambient temperature at the corresponding monitoring time. In this embodiment, the electronic incense burner is controlled by PID using reference P, I, and D parameters during the period from the most recent power-on start-up to the initial temperature rise monitoring time. That is, the controller output signal during this period is generated by the reference P, I, and D parameters. The reference P, I, and D parameters can be calibrated through a standard load experiment and are usually pre-set at the factory. Monitoring continues from the initial temperature rise monitoring time, and when the cumulative monitoring time reaches a preset first time threshold, feature extraction begins. The first monitoring time to be analyzed is recorded as the time when the cumulative monitoring duration first reaches a preset first duration threshold, starting from the initial temperature rise monitoring time. All monitoring times that are later than the first monitoring time and belong to the same run or work cycle are also recorded as monitoring times to be analyzed. In this embodiment, PID temperature control is performed using reference P and D parameters from the initial temperature rise monitoring time to the first monitoring time to be analyzed. That is, no feature extraction is performed during this stage, and the controller output signal during this stage is generated by the reference P and D parameters. Integration is used to eliminate... Steady-state error: Even if the proportional action has brought the temperature close to the target, small residual deviations (such as temperature differences caused by heat dissipation) will accumulate through the integral term, gradually adjusting the control output until the deviation is zero. However, the integral action may also lead to integral saturation, especially when the deviation is large. Therefore, in this embodiment, the integral is not used in the heating rate increment stage; only P and D are used for control. The stage from the initial heating monitoring time to the first monitoring time to be analyzed belongs to the initial heating increment stage, so only P and D are used for control thereas. The P parameter represents the proportional parameter, I represents the integral parameter, and D represents the derivative parameter. In specific applications, the implementer needs to set a preset temperature threshold according to the actual situation. For example, in this embodiment, the preset temperature threshold can be set to 2 degrees Celsius.

[0020] In practical applications, implementers need to set a preset first duration threshold according to the actual situation. For example, in this embodiment, the preset first duration threshold can be set to 60 seconds, that is, the monitoring time when the cumulative monitoring time first reaches 60 seconds from the initial heating monitoring time is the first monitoring time to be analyzed. In this embodiment, the sensor for collecting the real-time temperature of the furnace is deployed at the center of the heating plate, and the sensor for collecting the ambient temperature needs to be deployed close to the location of the electronic incense burner, such as on the side wall of the electronic incense burner. In this embodiment, the sensor is a high-precision NTC thermistor sensor. The sensor in this embodiment collects data synchronously, and the acquisition frequency is set according to the actual situation. For example, the acquisition frequency can be set to 2 Hz, and the resolution can be set to 0.1 degrees Celsius. In practical applications, implementers need to set a preset temperature threshold according to the actual situation. For example, in this embodiment, the preset temperature threshold can be set to 2 degrees Celsius.

[0021] Next, a local time window is obtained at the monitoring time to be analyzed. Each local time window at the monitoring time to be analyzed consists of the monitoring time to be analyzed and the historical monitoring times preceding it. In this embodiment, the length of the local time window at the monitoring time to be analyzed is set to a preset first duration threshold. Then, the local time windows at the monitoring time to be analyzed are normalized. The original furnace temperature at each monitoring time within the local time window is first filtered and then normalized. The normalized local time window is recorded as the normalized time window, the normalized monitoring time is recorded as the normalized monitoring time, the filtered original furnace temperature is recorded as the filtered temperature, and the normalized filtered temperature is recorded as the normalized furnace temperature. Therefore, the normalized furnace temperature corresponding to each normalized monitoring time within the normalized time window at the monitoring time to be analyzed can be obtained. Normalizing the local time window compresses its duration to a range of 0 to 1. For example, the expression for the normalized monitoring time obtained after normalizing any monitoring time within any local time window is: , For this monitoring time, This is the start time of the local window. This is the end time of the local window. The time interval of this local time window is defined as the linear normalization process performed on the local time window. The normalized furnace temperature is obtained by normalizing the filtered temperature at any monitoring moment within any local time window. T0 is the filtered temperature at that monitoring moment. This refers to the filtered temperature at the start monitoring time within this local time window. This refers to the filtered temperature at the end of the monitoring period within this local time window. Since it is in the heating phase, therefore... The minimum filtered temperature within this local time window, The maximum filtered temperature within this local time window is also the linear normalization. The above filtering refers to the filtering of the original furnace temperature within the local time window using a median filter with a window width of 5 to eliminate random pulse noise during the acquisition process. The median filtering process is a well-known technique.

[0022] Therefore, this embodiment can obtain the normalized furnace temperature corresponding to each normalized monitoring time in the normalized time window of the electronic incense burner under the time to be analyzed and monitored through the above process.

[0023] Step S002: Based on the normalized monitoring time and normalized furnace temperature, obtain the local convexity coefficient and local thermal inertia coefficient at the monitoring time to be analyzed, and perform asymptotic recursive global estimation on the local convexity coefficient and local thermal inertia coefficient to obtain the global convexity coefficient and global thermal inertia coefficient at different monitoring times to be analyzed. Record the variances of the local convexity coefficient and local thermal inertia coefficient at the first t monitoring times to be analyzed as the variances of the convexity coefficient and thermal inertia coefficient at the t-th monitoring time to be analyzed, respectively. The global convexity coefficient and global thermal inertia coefficient are used to adjust the baseline P, I, and D parameters to obtain the undetermined P, I, and D parameters at the t-th monitoring time to be analyzed. Based on the original furnace temperature in the neighborhood of the t-th monitoring time to be analyzed, as well as the variance of the convexity coefficient and the variance of the thermal inertia coefficient, it is determined whether the target inflection point and characteristic convergence point occur at the t-th monitoring time to be analyzed. Based on the determination result and the undetermined P, I, and D parameters, the controller output signal at the t-th monitoring time to be analyzed is adaptively generated and driven to execute.

[0024] This embodiment primarily achieves adaptive PID temperature control of the electronic incense burner through local feature extraction, progressive global estimation, convergence, and inflection point discrimination. Specifically, based on the determined time window, the local convexity coefficient and local thermal inertia coefficient are first evaluated. The convexity coefficient and thermal inertia index can quantify the normalized furnace temperature curve shape. Then, progressive global estimation is performed to obtain the global convexity coefficient and global thermal inertia coefficient. Based on the global convexity coefficient and global thermal inertia coefficient, feature convergence discrimination is performed, and the target inflection point is also discriminated, thereby obtaining the target inflection point time and feature convergence time. Based on the discrimination results and the extracted global convexity coefficient and global thermal inertia coefficient, adaptive PID temperature control is performed to ensure the effectiveness of PID temperature control for the electronic incense burner.

[0025] Therefore, this embodiment will first perform local feature analysis based on the normalized furnace temperature corresponding to each normalized monitoring time within the normalized time window. In other words, this embodiment will first obtain the local convexity coefficient and local thermal inertia coefficient at each monitoring time within the normalized time window to be analyzed, based on the normalized furnace temperature corresponding to each monitoring time within the normalized time window to be analyzed. Since the process of obtaining the local convexity coefficient and local thermal inertia coefficient at different monitoring times to be analyzed is the same, for ease of understanding and description, this embodiment will subsequently describe the process of obtaining the local convexity coefficient and local thermal inertia coefficient at the t-th monitoring time to be analyzed as an example. That is, the process of obtaining the local convexity coefficient and local thermal inertia coefficient at the t-th monitoring time to be analyzed is as follows:

[0026] First, obtain the second derivative at each normalized monitoring time within the normalized time window under the t-th monitoring time to be analyzed; the expression for the second derivative at the i-th normalized monitoring time within the normalized time window under the t-th monitoring time to be analyzed is: , and Let be the first derivatives at the (i+1)th and (i-1)th normalized monitoring times within the normalized time window under the t-th monitoring time to be analyzed. The normalized time step is the time interval between adjacent normalized monitoring moments within the normalized time window, or the absolute value of the difference between adjacent normalized monitoring moments within the normalized time window. Alternatively, as another real-time method, it can also be... As the second derivative at the i-th normalized monitoring time within the normalized time window under the t-th monitoring time to be analyzed. , and Let be the normalized furnace temperature corresponding to the (i-1), ith, and (i+1)th normalized monitoring times within the normalized time window under the t-th monitoring time to be analyzed, where i is not equal to 1 or N, and N is the total number of normalized monitoring times within the normalized time window under the t-th monitoring time to be analyzed; and let be the first derivative corresponding to the ith normalized monitoring time. The expression is , and Similarly, the first derivative at the i-th normalized monitoring time within the normalized time window under the t-th monitoring time to be analyzed is determined using the central difference method. The first derivatives at the two ends of the normalized monitoring time within the normalized time window under the t-th monitoring time to be analyzed can be determined using forward differencing and backward differencing, respectively. That is, the second derivative is calculated using forward differencing for the initial normalized monitoring time within the normalized time window, and using backward differencing for the final normalized monitoring time within the normalized time window. The expression for the first derivative at the i-th normalized monitoring time is: The expression for the first derivative at the first normalized monitoring time within the normalized time window at the t-th monitoring time to be analyzed is: The first derivative expression for the last normalized monitoring time in the normalized time window at the t-th monitoring time to be analyzed is: , and Let be the normalized furnace temperature corresponding to the first and second normalized monitoring times within the normalized time window under the t-th monitoring time to be analyzed. and Let t be the normalized furnace temperature corresponding to the last and second-to-last normalized monitoring times within the normalized time window under the t-th monitoring time to be analyzed; the above second derivative can reflect the rate of change of the heating rate and is dimensionless. When the first derivative is large, it indicates that the temperature rises rapidly, usually in the early stage of heating (such as the early stage of heating of small heat capacity incense materials) or the later stage (such as the acceleration stage of large heat capacity incense materials). When the first derivative is small, it indicates that the heating is slow, usually in the early stage of heating (such as the stage of overcoming thermal inertia of large heat capacity incense materials).

[0027] Additionally, it should be noted that the heating curve of an electronic incense burner directly reflects the heat capacity characteristics of the incense materials. For example, for incense materials with high heat capacity (such as sandalwood chips or large quantities), a large amount of heat is absorbed in the initial stage of heating, but the temperature rises slowly. As the temperature increases, the heat conduction efficiency increases, and the heating rate gradually accelerates, forming a downward convex curve (the second derivative is always positive). For incense materials with low heat capacity (such as essential oils or small quantities), the opposite is true: the temperature rises rapidly in the initial stage and then slows down in the later stage, forming an upward convex curve (the second derivative is always negative). Therefore, if the monitoring time is at... A second derivative greater than 0 indicates an increasing heating rate, resulting in a convex downward-pointing normalized furnace temperature curve. This also suggests a large system heat capacity, with improved heat transfer efficiency and faster heating as temperature rises – a typical characteristic of high-heat-capacity incense materials like sandalwood chips. Conversely, a second derivative less than 0 at the monitoring time indicates a decreasing heating rate, resulting in a convex upward-pointing normalized furnace temperature curve. This suggests a continuously decreasing heating rate, a smaller system heat capacity, rapid initial heating, and a subsequent decrease in temperature as the target temperature approaches or the incense material volatilizes. The gradual change in temperature is a typical characteristic of essential oils and other low-heat-capacity fragrance materials. The second derivative at the monitoring point equals 0, indicating a near-linear relationship. This marks the turning point where the heating rate decreases, signifying the transition from inertia-driven to response-driven phases. The sign and magnitude of the second derivative directly reflect the changing trend of the fragrance material's thermal inertia. Therefore, in this embodiment, the overall curvature direction of the curve within the window will be quantified by weighted integration of the second derivative within the window—that is, the local convexity coefficient. A positive local convexity coefficient corresponds to downward convexity, and a negative local convexity coefficient corresponds to upward convexity. The absolute value of the local convexity coefficient reflects the severity of the curvature. Therefore, the convexity coefficient directly determines the orientation of the PID control strategy. For high-heat-capacity fragrance materials or downward convex curves, a stronger proportional action is needed to overcome initial thermal inertia; for upward convex curves or low-heat-capacity fragrance materials, a stronger differential action is needed to prevent overshoot later. Therefore, the convexity coefficient is the primary basis for adaptive parameter adjustment. The system in this embodiment is an intelligent temperature control system for an electronic incense burner.

[0028] Following this, this embodiment obtains the local convexity coefficient at the t-th monitoring time based on the second derivative, the normalized monitoring time, and the normalized time step size within the normalized time window under the t-th monitoring time to be analyzed; and the expression for the local convexity coefficient at the t-th monitoring time to be analyzed is:

[0029]

[0030] in, Let be the local convexity coefficient at the t-th monitoring time to be analyzed. Let be the second derivative at the i-th normalized monitoring time within the normalized time window under the t-th monitoring time to be analyzed. Let i be the i-th normalized monitoring time within the normalized time window at the t-th monitoring time to be analyzed. Here, e is a preset constant, and e is a natural constant. This is a weighting function used to emphasize the morphological features of the early stages of the window; that is, the earlier the window is, the greater the weight. To prevent the denominator from being 0, this embodiment sets it to... Furthermore, when the curve formed by the normalized furnace temperature corresponding to the normalized monitoring time within the normalized time window at the t-th monitoring time to be analyzed exhibits a convex downward shape, indicating that the second derivative is smaller or the heating is slower in the early stage of the window, and larger or the heating is faster in the later stage, then after weighted integration... A positive value indicates that when the time series curve of the normalized furnace temperature corresponding to the normalized monitoring time within the normalized time window at the t-th monitoring time to be analyzed exhibits a convex downward shape, the second derivative is larger or the temperature rises faster in the early stage of the window, and smaller or the temperature rises slower in the later stage. Therefore, after weighted integration... If the value is negative, the time series curve of the normalized furnace temperature corresponding to the normalized monitoring time in the normalized time window under the t-th monitoring time to be analyzed will have an upward convex shape.

[0031] Because thermal inertia can cause problems such as response lag, increased overshoot risk, and decreased stability in temperature control systems, thermal inertia must also be considered during regulation. Therefore, in this embodiment, the local thermal inertia coefficient at the t-th monitoring time will be obtained based on each normalized monitoring time, the corresponding normalized temperature, and the normalized time step within the normalized time window at the t-th monitoring time. The expression for the local thermal inertia coefficient at the t-th monitoring time is as follows:

[0032]

[0033] in, Let be the local thermal inertia coefficient at the t-th time point to be analyzed and monitored. Let be the normalized furnace temperature corresponding to the j-th normalized monitoring time within the normalized time window under the t-th monitoring time to be analyzed. Let j be the j-th normalized monitoring time within the normalized time window at the t-th monitoring time to be analyzed. Greater than This indicates that the actual furnace temperature here is higher than the expected value for uniform heating, and also indicates that the actual heating is faster than the ideal line. The ideal line represents uniform heating, and this is considered a case of heating ahead of schedule. Less than When this occurs, it indicates that the actual temperature rise lags behind the expected value of uniform temperature rise, which is ideal. It also indicates that the actual temperature rise lags behind the ideal line. This is considered a temperature lag, or in other words, when... If the value is higher than the ideal line or higher than the expected value for uniform heating, it indicates that the heating is ahead of schedule. If the value is lower than the ideal line or lower than the expected value for uniform heating, it indicates that the heating is lagging; therefore, it can be concluded that when... When the value is positive, it indicates that the actual temperature rise is faster than the ideal curve, exhibiting a temperature lead characteristic. Therefore, this means that the actual furnace temperature at this point is already higher than the expected value for uniform temperature rise. When the value is positive, it indicates that the t-th moment to be analyzed and monitored is more likely to be in a state of advanced temperature rise. A negative value indicates that the actual temperature rise lags behind the ideal line, exhibiting a temperature lag characteristic. This suggests that the t-th monitoring time is more likely to be in a state of temperature lag. When the value is 0, the heating rate is uniform. The ideal line is in the normalized two-dimensional coordinate system from the coordinate... To coordinates The diagonal of the normalized two-dimensional coordinate system represents the normalized time on the horizontal axis and the normalized temperature on the vertical axis. That is, the horizontal axis ranges from 0 to 1 and the vertical axis ranges from 0 to 1. The larger the absolute value, the more the actual temperature rise deviates from the ideal line, indicating a higher thermal inertia.

[0034] After obtaining the local convexity coefficient and local thermal inertia coefficient at the time to be analyzed and monitored, in order to reflect the long-term thermal response essence of the material and avoid frequent oscillations in the integral gain caused by local fluctuations, so as to achieve stable and smooth temperature control, it is necessary to perform an asymptotic recursive global estimation of the local convexity coefficient and local thermal inertia coefficient at the time to be analyzed and monitored, to obtain the global convexity coefficient and global thermal inertia coefficient at the time to be analyzed and monitored; subsequently, the global convexity coefficient and global thermal inertia coefficient are used for parameter adjustment; and the method for obtaining the local convexity coefficient and local thermal inertia coefficient at any time to be analyzed and monitored in this embodiment is the same as the method for obtaining the local convexity coefficient and local thermal inertia coefficient at the t-th time to be analyzed and monitored mentioned above; and this embodiment does not perform analysis on the time to be analyzed and monitored during the initial heating stage. Global estimation begins from the monitoring times for which the number of historical monitoring times to be analyzed is not less than a preset threshold. That is, it starts from the (B+1)th monitoring time to be analyzed and performs a progressive recursive global estimation, where B is the preset threshold. If the current monitoring time to be analyzed is the tth monitoring time, and the number of monitoring times to be analyzed before the tth monitoring time is not less than the preset threshold, then this embodiment will describe the process of progressively recursively estimating the local convexity coefficient and local thermal inertia coefficient at the tth monitoring time to obtain the global convexity coefficient and global thermal inertia coefficient at the tth monitoring time as an example. Therefore, the specific process of obtaining the global convexity coefficient and global thermal inertia coefficient at the tth monitoring time is as follows:

[0035] Calculate the mean of the local convexity coefficients for the first t monitoring times to be analyzed, and record it as the global convexity coefficient for the t-th monitoring time to be analyzed. Calculate the mean of the local thermal inertia coefficients for the first t monitoring times to be analyzed, and record it as the global thermal inertia coefficient for the t-th monitoring time to be analyzed. The global convexity coefficient and global thermal inertia coefficient are mainly used for adaptive adjustment of PID parameters. In specific applications, implementers can set preset quantity thresholds according to actual conditions, such as setting the preset quantity threshold B to 5. When the global thermal inertia coefficient at the t-th monitoring time is positive, it indicates that the t-th monitoring time is more likely to be in a state of temperature lead. In this case, the proportional parameter should be reduced to suppress system overshoot, avoid thermal stress damage, improve temperature control accuracy and process stability, or the control output intensity should be reduced to make the heating power increase more slowly and offset the thermal inertia effect. When the global thermal inertia coefficient at the t-th monitoring time is negative, it indicates that the t-th monitoring time is more likely to be in a state of temperature lag. The proportional parameter should be further increased to compensate for system thermal inertia and response delay, improve control dynamic performance, or the control output intensity should be increased to make the heating power increase more rapidly and offset the thermal inertia effect. When the absolute value of the global thermal inertia coefficient at the t-th monitoring time is larger, it indicates that the actual normalized furnace temperature deviates more from the baseline at this time. Therefore, the integral should be strengthened to... To eliminate steady-state errors and ensure the quality of aroma release, when the global convexity coefficient at the t-th time point under analysis is positive, it indicates that the system's thermal response is convex downwards, or the normalized furnace temperature time series curve is convex downwards. This is characterized by slow initial heating, accelerated later heating, sufficient material heat storage, and gradual energy release. In this case, increasing the proportional parameter can improve response sensitivity and avoid temperature hysteresis, while increasing the differential parameter can suppress later overshoot or overadjustment in advance, preventing aroma charring. When the global convexity coefficient at the t-th time point under analysis is negative, it indicates that the system's thermal response is convex upwards, or the normalized furnace temperature time series curve is convex upwards. This is characterized by excessive initial heating, weak later heating, premature exhaustion of material heat energy, and premature aroma release with a hint of burnt smell. In this case, it is necessary to reduce the proportional parameter to suppress overreaction and reduce the differential parameter to prevent temperature control oscillations and local overheating.

[0036] Next, the variance of the local convexity coefficient under the first t monitoring times to be analyzed is calculated and denoted as the variance of the convexity coefficient under the t-th monitoring time. The variance of the local thermal inertia coefficient under the first t monitoring times to be analyzed is also calculated and denoted as the variance of the thermal inertia coefficient under the t-th monitoring time. The variances of the convexity coefficient and the thermal inertia coefficient are mainly used to determine the convergence of features. When the features converge, it indicates that the system can rely on the current global features, and it also indicates that the system's estimation of the convexity and thermal inertia of the current incense material has stabilized. The subsequent local features no longer fluctuate significantly, and these feature parameters can be safely used for adaptive tuning of PID parameters. In other words, when the features converge, the subsequent PID parameters no longer change, and the PID parameters obtained when the features converge are used for temperature control until the electronic incense burner is turned off or powered off, or until the current operation or current working cycle of the electronic incense burner ends.

[0037] If the current monitoring time to be analyzed is the t-th monitoring time, and the number of monitoring times to be analyzed before the t-th monitoring time is not less than a preset threshold, then this embodiment will subsequently describe the process of adaptive PID temperature control or adaptive generation of the controller output signal at the t-th monitoring time as an example. Therefore, after obtaining the global convexity coefficient and global thermal inertia coefficient at the t-th monitoring time, the reference P, I, and D parameters are adjusted based on the global convexity coefficient and global thermal inertia coefficient at the t-th monitoring time to obtain the P, I, and D parameters to be determined at the t-th monitoring time. The expression for the P, I, and D parameters to be determined at the t-th monitoring time is:

[0038]

[0039]

[0040]

[0041] in, Let P be the parameter to be determined at the t-th monitoring time. Let I be the parameter to be determined at the t-th monitoring time. Let KP0, KI0, and KD0 be the parameters to be determined at the t-th monitoring time, and let KP0, KI0, and KD0 be the baseline P parameter, baseline I parameter, and baseline D parameter, respectively. , , and These are the preset first feature coefficient, the preset second feature coefficient, the preset third feature coefficient, and the preset fourth feature coefficient, respectively. Let be the global convexity coefficient at the t-th monitoring time to be analyzed. Let be the global thermal inertia coefficient at the t-th time point to be analyzed and monitored; in specific applications, the implementer can set the characteristic coefficient according to actual conditions such as experimental calibration and fitting, as in this embodiment, it can be set to... , , , .

[0042] And if A value greater than 0 indicates that the system's thermal response exhibits a convex downward sloping shape, or that the normalized furnace temperature time-series curve at this point also exhibits a convex downward sloping shape. Further enhancement of the baseline P and D parameters, or enhancement of the proportional and derivative parameters, is required. When greater than 0, Greater than 1 and A value greater than 1 can enhance both the proportional and differential parameters, i.e. Greater than 0 and the larger the value, and The larger; if A value less than 0 indicates that the system's thermal response exhibits an upward convex shape, or that the normalized furnace temperature time-series curve at this point also exhibits an upward convex shape. Therefore, it is necessary to further reduce the baseline P and D parameters, or to reduce the proportional and derivative parameters. When less than 0, Less than 1 and A value less than 1 allows for a reduction in both the proportional and derivative parameters, i.e. Less than 0 and the smaller the value, and The smaller. If A positive value indicates that the t-th moment under analysis is more likely to be in a state of temperature lead. To suppress system overshoot and counteract the effects of thermal inertia, the proportional parameter needs to be reduced or further reduced based on the baseline proportional parameter. For positive, Less than 0, therefore a reduction in the scaling factor can be achieved, that is... When it is positive and larger, The smaller; if A negative value indicates that the t-th monitoring time is more likely to be in a state of temperature lag. To quickly respond to the temperature lag gap, the proportional parameter needs to be further increased, or it needs to be further increased based on the baseline proportional parameter. When it is negative, A value greater than 0 allows for an increase in the scaling factor, i.e. The smaller the value, the more negative the value. The larger. The larger the absolute value, the more the actual normalized furnace temperature deviates from the baseline. Therefore, the integration should be strengthened to eliminate steady-state errors. The larger the absolute value, The larger, The smaller the absolute value, The smaller.

[0043] In this embodiment, after obtaining the P, I, and D parameters to be determined at the t-th time to be analyzed and monitored, the system determines whether the target inflection point and characteristic convergence point have occurred at the t-th time to be analyzed and monitored based on the original furnace temperature in the neighborhood of the t-th time to be analyzed and monitored, as well as the variance of the convexity coefficient and the variance of the thermal inertia coefficient at the t-th time to be analyzed and monitored. Based on the determination result and the P, I, and D parameters to be determined at the t-th time to be analyzed and monitored, the system generates the controller output signal at the t-th time to be analyzed and monitored and drives its execution. That is, the actuator drives the output controller signal to execute, so as to realize the temperature control of the electronic incense burner. The actuator is an internal component of the PID temperature control system of the electronic incense burner.

[0044] In this embodiment, the process of determining whether the target inflection point and feature convergence point occur at the t-th monitoring time and generating the controller output signal at the t-th monitoring time based on the determination result and the undetermined P, I, and D parameters at the t-th monitoring time is as follows:

[0045] First, the target inflection point and feature convergence point are identified sequentially. If neither is identified before the t-th monitoring time, the identification process continues at the t-th monitoring time. If neither is identified at the t-th monitoring time, the controller output signal for the t-th monitoring time is generated based on the preset reference weights and undetermined P and D parameters. If the target inflection point is still not identified at the t-th monitoring time but the feature convergence point has been identified, the controller output signal for the t-th monitoring time is generated based on the undetermined P and D parameters at the feature convergence point; that is, feature convergence is directly achieved at this point. The controller output signal generated by the undetermined P and D parameters at time t is used as the controller output signal at the t-th time to be analyzed and monitored. If the target inflection point time has been identified but the feature convergence time has not been identified at the t-th time to be analyzed and monitored, the controller output signal at the t-th time to be analyzed and monitored is generated based on the preset reference weights and the undetermined P, I, and D parameters at the t-th time to be analyzed and monitored. If both the target inflection point time and the feature convergence time are identified at the t-th time to be analyzed and monitored, the controller output signal at the t-th time to be analyzed and monitored is generated based on the undetermined P, I, and D parameters at the feature convergence time. That is, at this time, the controller output signal generated by the undetermined P, I, and D parameters at the feature convergence time is directly used as the controller output signal at the t-th time to be analyzed and monitored.

[0046] If the target inflection point and feature convergence point have been identified before the t-th monitoring time to be analyzed, then the controller output signal at the t-th monitoring time to be analyzed is directly generated based on the undetermined P parameters, I and D parameters at the feature convergence point. In other words, the controller output signal generated based on the undetermined P parameters, I and D parameters at the feature convergence point is directly used as the controller output signal at the t-th monitoring time to be analyzed.

[0047] If the target inflection point has been identified before the t-th monitoring time but the feature convergence time has not been identified, then only the feature convergence time will be identified at the t-th monitoring time. If the feature convergence time has still not been identified at the t-th monitoring time, then the controller output signal for the t-th monitoring time will be generated based on the preset reference weights and the P, I, and D parameters to be determined at the t-th monitoring time. , and The generated control output signal is used as the controller output signal at the t-th monitoring time to be analyzed. If a characteristic convergence time is identified at the t-th monitoring time, the controller output signal at the t-th monitoring time is generated based on the undetermined P, I, and D parameters at the characteristic convergence time. That is, the controller output signal generated based on the undetermined P, I, and D parameters at the characteristic convergence time is directly used as the controller output signal at the t-th monitoring time. If the target inflection point time is not identified before the t-th monitoring time but a characteristic convergence time has been identified, then only the target inflection point time is identified at the t-th monitoring time. If the target inflection point time is still not identified at the t-th monitoring time, the controller output signal at the t-th monitoring time is generated based on the undetermined P and D parameters at the characteristic convergence time. If the target inflection point time is identified at the t-th monitoring time, the controller output signal at the t-th monitoring time is generated based on the undetermined P, I, and D parameters at the characteristic convergence time.

[0048] Generating the controller output signal at the t-th monitoring time based on the preset reference weights and undetermined P and D parameters at the t-th monitoring time refers to generating the controller output signal at the t-th monitoring time based on the parameters in parameter combination 1; generating the controller output signal at the t-th monitoring time based on the undetermined P and D parameters at the characteristic convergence time refers to generating the controller output signal at the t-th monitoring time based on the parameters in parameter combination 2; generating the controller output signal at the t-th monitoring time based on the preset reference weights and undetermined P, I, and D parameters at the t-th monitoring time refers to generating the controller output signal at the t-th monitoring time based on the parameters in parameter combination 3; generating the controller output signal at the t-th monitoring time based on the undetermined P, I, and D parameters at the characteristic convergence time refers to generating the controller output signal at the t-th monitoring time based on the parameters in parameter combination 4; parameter combination 1 is... and The parameter combination 2 consists of the undetermined P and D parameters at the characteristic convergence time, and the parameter combination 3 consists of... , and The parameter combination 4 is composed of the undetermined P, I, and D parameters at the characteristic convergence time. , and These are the adjusted weighted P, I, and D parameters at the t-th monitoring time to be analyzed; the process of generating the controller output signal using known P, I, and D parameters is a well-known technique.

[0049] The expressions for the adjusted weighted P, I, and D parameters at the t-th monitoring time to be analyzed are:

[0050] ;

[0051] ;

[0052] ;

[0053] in, , and These are the adjusted weighted P, I, and D parameters at the t-th monitoring time to be analyzed. Let be the preset reference weights for the t-th monitoring time to be analyzed. , and Let P, I, and D be the parameters to be determined at the t-th monitoring time to be analyzed.

[0054] The expression for the preset reference weight at the t-th monitoring time to be analyzed is:

[0055]

[0056] in, Here, the preset reference weights are defined for the t-th monitoring time to be analyzed, and Max() is the function to retrieve the maximum value. The standard deviation of all global convexity coefficients actually calculated during the time period from the first monitoring time to the t-th monitoring time is given. The standard deviation of all global thermal inertia coefficients actually calculated during the time interval from the first monitoring time to the t-th monitoring time; when and The smaller the value, or the closer it is to 0, the more reliable the global convexity coefficient and global thermal inertia coefficient at the t-th monitoring time. Therefore, when determining and adjusting the weighted P, I, and D parameters, more reference should be made to the PID parameters to be determined. and The larger the value, the less reliable the global convexity coefficient and global thermal inertia coefficient at the t-th monitoring time. Therefore, when determining and adjusting the weighted P, I, and D parameters, the baseline PID parameters should be referenced more. The larger the value, the greater the weight of the reference PID parameter to be determined when subsequently adjusting the weighted P, I, and D parameters; conversely, the greater the weight of the reference PID parameter, in order to further ensure the control effect. In addition, for the first occurrence of a historical monitoring time to be analyzed with a number of times not less than the preset number threshold, that is, the B+1th monitoring time to be analyzed, no global estimation was performed for the times before the B+1th monitoring time to be analyzed. Therefore, there is no global convexity coefficient and global thermal inertia coefficient before the B+1th monitoring time to be analyzed. In this case, the preset reference weight under the B+1th monitoring time to be analyzed is not calculated, and it can be directly set to 1.

[0057] In this embodiment, the specific process of identifying the target inflection point and feature convergence point in chronological order is as follows:

[0058] The time-series curve formed by the original furnace temperature at all monitoring times within the preset second time-threshold before the monitoring time to be analyzed is the original curve for the corresponding monitoring time to be analyzed. Points on the original curve are called original points. Cubic spline interpolation is performed on the original curve for the monitoring time to be analyzed, and the interpolated curve is recorded as the interpolated curve for the corresponding monitoring time to be analyzed. Cubic spline interpolation is a well-known technique. The original curve for any monitoring time to be analyzed is the curve obtained by mapping the original coordinate system to all monitoring times and the original furnace temperature at the monitoring time within the preset second time-threshold before the monitoring time to be analyzed, and connecting the points obtained by mapping in chronological order. The horizontal axis of the original coordinate system represents unnormalized time, and the vertical axis represents unnormalized temperature. In specific applications, the implementer needs to set the preset second time-threshold according to the actual situation, such as setting the preset second time-threshold to 120 seconds. If the number of monitoring times to be analyzed before the w-th monitoring time is equal to a preset threshold, then starting from the w-th monitoring time, inflection points are identified sequentially on the interpolation curves of the monitoring times to be analyzed. The monitoring time to be analyzed where an inflection point is first identified is recorded as the re-analysis monitoring time, and the inflection point on the interpolation curve of the identified re-analysis monitoring time is recorded as the theoretical inflection point. The process continues to determine whether there is an inflection point on the original curve of the re-analysis monitoring time. If there is, the inflection point on the original curve of the re-analysis monitoring time that is closest to the theoretical inflection point is selected as the target inflection point. If there is no inflection point, the original curve of the re-analysis monitoring time with the closest comprehensive distance to the theoretical inflection point is selected. Points are designated as target inflection points, and the time corresponding to these inflection points is recorded as the target inflection point time. This means the specific time value of the target inflection point on the time axis is the target inflection point time, indicating that the target inflection point time was identified at the reanalysis monitoring time. It also indicates that the target inflection point time was identified before the monitoring time following the reanalysis monitoring time, therefore, further identification of target inflection points is unnecessary, and the identification process stops. Points on the curve where the second derivative is 0 are inflection points. The weighted sum of the normalized distance of the second derivative and the normalized distance of the position coordinates between any original point on the original curve and the theoretical inflection point is the comprehensive distance between the corresponding original point and the theoretical inflection point. The expression for the comprehensive distance between any original point and the theoretical inflection point is: ,in, and The first and second weighting factors are defined separately. Norm() is the linear normalization function. D1 is the Euclidean distance between the second derivative of the original point on the original curve and the second derivative of the theoretical inflection point. D2 is the Euclidean distance between the position coordinates of the original point on the original curve and the position coordinates of the theoretical inflection point. The second derivative of the theoretical inflection point is 0. In specific applications, implementers can set these parameters according to the actual situation. and If it can be set 0.4 The value is 0.6. For example: When identifying inflection points on the interpolation curve at the w-th monitoring time, if no inflection point exists on the interpolation curve at the w+1-th monitoring time, the process continues. If an inflection point exists on the interpolation curve at the w-th monitoring time, this inflection point is recorded as the theoretical inflection point. The process then continues to determine if an inflection point exists on the original curve at the w+1-th monitoring time. If it exists, all inflection points on the original curve at the w+1-th monitoring time are recorded as candidate inflection points, and the candidate inflection point closest to the theoretical inflection point in terms of coordinate position is selected as the target inflection point. If no inflection point exists, the original point on the original curve at the w+1-th monitoring time with the closest comprehensive distance to the theoretical inflection point is selected as the target inflection point.

[0059] Simultaneously, it is determined whether the variances of the convexity coefficient and thermal inertia coefficient at the w-th monitoring time to be analyzed are both less than a preset variance threshold. If they are not both less than the preset variance threshold, it is then determined whether the variances of the convexity coefficient and thermal inertia coefficient at the (w+1)-th monitoring time to be analyzed are both less than the preset variance threshold. If they are both less than the preset variance threshold, the (w+1)-th monitoring time to be analyzed is recorded as the characteristic convergence time, indicating that a characteristic convergence time has been identified at the (w+1)-th monitoring time to be analyzed, and also indicating that a characteristic convergence time has been identified before other monitoring times after the (w+1)-th monitoring time to be analyzed. Therefore, there is no need to identify characteristic convergence times again, i.e., the identification of characteristic convergence times is stopped. In specific applications, the preset variance threshold can be set according to the actual situation. For example, in PID parameter adjustment, an adjustment of less than 5% will only cause a small performance change in actual control and will not cause system instability or significant deterioration of control effect. Therefore, in this embodiment, the preset variance threshold can be set to 0.05.

[0060] In this embodiment, the method for obtaining the undetermined P, I, and D parameters at the characteristic convergence time is the same as the method for obtaining the undetermined P, I, and D parameters at the t-th analysis and monitoring time. The only difference is that the global convexity coefficient and global thermal inertia coefficient at the t-th analysis and monitoring time in the expression of the undetermined P, I, and D parameters at the t-th analysis and monitoring time are replaced with the global convexity coefficient and global thermal inertia coefficient at the characteristic convergence time.

[0061] Since the integral parameter is mainly used to eliminate steady-state error, and even if the proportional action has brought the temperature close to the target, small residual deviations (such as temperature differences caused by heat dissipation) will accumulate through the integral term, gradually adjusting the control output until the deviation is zero. However, the integral action may also lead to integral saturation, especially when the deviation is large. Therefore, this embodiment sets the temperature rise rate increasing stage, that is, the stage from the initial temperature rise monitoring time to the target inflection point time. At this time, the target inflection point time is not included, and only PD control is performed. In this stage, the system mainly overcomes thermal inertia and requires a strong proportional action. In addition, the monitoring time of the global convexity coefficient and the global thermal inertia coefficient is not calculated in the stage from the initial temperature rise monitoring time to the target inflection point time. Instead, the controller output signal at the corresponding monitoring time is generated based on the reference P parameter and the reference D parameter and driven execution is performed. In this embodiment, global PID control is applied to the stage of decreasing heating rate, which is the stage when the target temperature is first reached at the target inflection point. During this stage, the system approaches the target temperature, requiring strong derivative action to suppress overshoot. Furthermore, if the absolute value of the difference between the original furnace temperature and the target temperature at a certain moment is less than a preset accuracy threshold, it indicates that the target temperature has been reached at that moment. In practical applications, the implementer can set the preset accuracy threshold according to the actual situation, such as setting it to 1 degree Celsius. Additionally, from the moment the target temperature is reached until the end of the current operation of the electronic incense burner, the controller output signal at the corresponding moment is generated based on the undetermined P, I, and D parameters at the feature convergence moment. The use of fixed PID parameters for temperature control when the feature has converged in this embodiment avoids parameter jitter caused by small fluctuations after feature stabilization, ensuring the determinism of control.

[0062] Thus, this embodiment completes the adaptive PID temperature control of the electronic incense burner. Moreover, this embodiment, through feature extraction, namely the extraction of convexity coefficient and thermal inertia coefficient, and based on the extracted features, as well as the identification results of inflection points and feature convergence, adaptively performs PID temperature control or adaptively generates controller output signals and drives execution. This can eliminate steady-state errors, reduce overshoot and oscillation, and improve temperature control accuracy, thereby improving the temperature control effect.

[0063] In summary, this embodiment first obtains the normalized furnace temperature corresponding to each normalized monitoring time within the normalized time window of the electronic incense burner under the monitoring time to be analyzed; then, based on the normalized monitoring time and normalized furnace temperature, it obtains the local convexity coefficient and local thermal inertia coefficient under the monitoring time to be analyzed, and performs a progressive recursive global estimation on the local convexity coefficient and local thermal inertia coefficient to obtain the global convexity coefficient and global thermal inertia coefficient under different monitoring times to be analyzed. The variance of the local convexity coefficient and the variance of the local thermal inertia coefficient under the first t monitoring times to be analyzed are respectively denoted as the t-th monitoring time to be analyzed. The variances of the convexity coefficient and thermal inertia coefficient at the measurement time are used to adjust the baseline P, I, and D parameters based on the global convexity coefficient and global thermal inertia coefficient, resulting in the undetermined P, I, and D parameters at the t-th monitoring time. Finally, based on the original furnace temperature in the neighborhood of the t-th monitoring time and the variances of the convexity coefficient and thermal inertia coefficient, it is determined whether the target inflection point and feature convergence point have occurred at the t-th monitoring time. Based on the determination results and the undetermined P, I, and D parameters, the controller output signal at the t-th monitoring time is adaptively generated and driven for execution. Furthermore, this embodiment, through the feature extraction of convexity coefficient and thermal inertia coefficient, and the integration of the feature extraction results, the target inflection point, and feature convergence point determination results into an adaptive PID temperature control strategy, can eliminate steady-state errors, reduce overshoot and oscillations, and improve temperature control accuracy. In other words, the adaptive PID temperature control strategy based on feature extraction results and combined with the target inflection point and feature convergence point determination results can improve temperature control performance.

[0064] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit them. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the spirit and scope of the technical solutions of the embodiments of this application, and should all be included within the protection scope of this application.

Claims

1. An artificial intelligence-based electronic incense burner adaptive PID temperature control method, characterized in that, The method includes the following steps: Obtain the normalized furnace temperature corresponding to each normalized monitoring time within the normalized time window of the electronic incense burner at the time to be analyzed and monitored; Based on the normalized monitoring time and normalized furnace temperature, the local convexity coefficient and local thermal inertia coefficient at the monitoring time to be analyzed are obtained. The local convexity coefficient and local thermal inertia coefficient are then estimated asymptotically to obtain the global convexity coefficient and global thermal inertia coefficient at different monitoring times to be analyzed. The variances of the local convexity coefficient and local thermal inertia coefficient at the first t monitoring times to be analyzed are denoted as the variances of the convexity coefficient and thermal inertia coefficient at the t-th monitoring time to be analyzed, respectively. Based on the global convexity coefficient and global thermal inertia coefficient, the baseline P, I, and D parameters are adjusted to obtain the P, I, and D parameters to be determined at the t-th monitoring time to be analyzed. Based on the original furnace temperature in the neighborhood of the t-th monitoring time and the variance of the convexity coefficient and the variance of the thermal inertia coefficient, it is determined whether the target inflection point and characteristic convergence time occur at the t-th monitoring time. Based on the determination result and the P, I and D parameters to be determined, the controller output signal at the t-th monitoring time is adaptively generated and driven to execute.

2. The artificial intelligence-based electronic incense burner adaptive PID temperature control method of claim 1, wherein, The methods for obtaining the local convexity coefficient and local thermal inertia coefficient at the t-th monitoring time to be analyzed include: Based on the second derivative at each normalized monitoring time within the normalized time window under the t-th monitoring time to be analyzed, the normalized monitoring time, and the normalized time step, the local convexity coefficient at the t-th monitoring time to be analyzed is obtained. Based on each normalized monitoring time within the normalized time window under the t-th monitoring time to be analyzed, the corresponding normalized furnace temperature, and the normalized time step, the local thermal inertia coefficient at the t-th monitoring time to be analyzed is obtained. The normalized time step is the time interval between adjacent normalized monitoring times within the normalized time window. The expression for the second derivative at the i-th normalized monitoring time within the normalized time window under the t-th monitoring time to be analyzed is: ,in, To normalize the time step, and Let i and n be the first derivatives at the (i+1)th and (i-1)th normalized monitoring times within the normalized time window under the t-th monitoring time to be analyzed, respectively, where i is not equal to 1 and N, and N is the total number of normalized monitoring times within the normalized time window under the t-th monitoring time to be analyzed.

3. The adaptive PID temperature control method for electronic incense burners based on artificial intelligence as described in claim 2, characterized in that, The expression for the local convexity coefficient at the t-th monitoring time to be analyzed is: ； in, Let be the local convexity coefficient at the t-th monitoring time to be analyzed. Let be the second derivative at the i-th normalized monitoring time within the normalized time window under the t-th monitoring time to be analyzed. Let i be the i-th normalized monitoring time within the normalized time window at the t-th monitoring time to be analyzed. is a preset constant, and e is a natural constant.

4. The adaptive PID temperature control method for electronic incense burners based on artificial intelligence as described in claim 2, characterized in that, The expression for the local thermal inertia coefficient at the t-th monitoring time to be analyzed is: ； in, Let be the local thermal inertia coefficient at the t-th monitoring time to be analyzed. Let be the normalized furnace temperature corresponding to the j-th normalized monitoring time within the normalized time window at the t-th monitoring time to be analyzed. Let j be the j-th normalized monitoring time within the normalized time window at the t-th monitoring time to be analyzed.

5. The adaptive PID temperature control method for electronic incense burners based on artificial intelligence as described in claim 1, characterized in that, The global convexity coefficient and global thermal inertia coefficient at the t-th monitoring time are the mean values ​​of the local convexity coefficient and local thermal inertia coefficient at the previous t monitoring times, respectively.

6. The adaptive PID temperature control method for electronic incense burners based on artificial intelligence as described in claim 1, characterized in that, The expressions for the undetermined P, I, and D parameters at the t-th monitoring time to be analyzed are: ； ； ； in, Let P be the parameter to be determined at the t-th monitoring time. Let I be the parameter to be determined at the t-th monitoring time. Let KP0, KI0, and KD0 be the parameters to be determined at the t-th monitoring time, and let KP0, KI0, and KD0 be the baseline P parameter, baseline I parameter, and baseline D parameter, respectively. , , and These are the preset first feature coefficient, the preset second feature coefficient, the preset third feature coefficient, and the preset fourth feature coefficient, respectively. Let be the global convexity coefficient at the t-th monitoring time to be analyzed. Let be the global thermal inertia coefficient at the t-th time point to be analyzed and monitored.

7. The adaptive PID temperature control method for electronic incense burners based on artificial intelligence as described in claim 6, characterized in that, A method for determining whether the target inflection point and feature convergence point occur at the t-th monitoring time to be analyzed, and adaptively generating the controller output signal at the t-th monitoring time based on the determination result and the P, I, and D parameters to be determined, includes: The target inflection point and feature convergence point are identified sequentially. If neither is identified before the t-th monitoring time, the identification continues at the t-th monitoring time. If neither is identified at the t-th monitoring time, the controller output signal for the t-th monitoring time is generated based on parameter combination 1. If the target inflection point is not identified at the t-th monitoring time but the feature convergence point is identified, the controller output signal for the t-th monitoring time is generated based on parameter combination 2. If the target inflection point is identified at the t-th monitoring time but the feature convergence point is not identified, the controller output signal for the t-th monitoring time is generated based on parameter combination 3. If both the target inflection point and feature convergence point are identified at the t-th monitoring time, the controller output signal for the t-th monitoring time is generated based on parameter combination 4. If the target inflection point and feature convergence point have been identified before the t-th monitoring time to be analyzed, then the controller output signal at the t-th monitoring time to be analyzed is generated based on parameter combination 4; if the target inflection point has been identified before the t-th monitoring time to be analyzed but the feature convergence point has not been identified, then the identification of the feature convergence point continues at the t-th monitoring time to be analyzed; if the feature convergence point has still not been identified at the t-th monitoring time to be analyzed, then the controller output signal at the t-th monitoring time to be analyzed is generated based on parameter combination 3; if the feature convergence point is identified at the t-th monitoring time to be analyzed, then the controller output signal at the t-th monitoring time to be analyzed is generated based on parameter combination 4. If the target inflection point is not identified before the t-th monitoring time to be analyzed, but the feature convergence time has been identified, then the identification of the target inflection point will continue at the t-th monitoring time to be analyzed. If the target inflection point is still not identified at the t-th monitoring time to be analyzed, then the controller output signal at the t-th monitoring time to be analyzed will be generated based on parameter combination 2. If the target inflection point is identified at the t-th monitoring time to be analyzed, then the controller output signal at the t-th monitoring time to be analyzed will be generated based on parameter combination 4. Parameter combination 1 is composed of and The parameter combination 2 consists of the undetermined P and D parameters at the characteristic convergence time, and the parameter combination 3 consists of... , and The parameter combination 4 is composed of the undetermined P, I, and D parameters at the characteristic convergence time. , and These are the adjusted weighted P, I, and D parameters for the t-th monitoring time to be analyzed.

8. The adaptive PID temperature control method for electronic incense burners based on artificial intelligence as described in claim 7, characterized in that, Methods for identifying the inflection point of a target include: If the number of monitoring times to be analyzed before the w-th monitoring time is equal to a preset threshold, then starting from the w-th monitoring time, inflection points are identified sequentially on the interpolation curves of the monitoring times to be analyzed. The monitoring time to be analyzed where an inflection point is first identified is recorded as the re-analysis monitoring time, and the inflection point on the interpolation curve of the identified re-analysis monitoring time is recorded as the theoretical inflection point. The process continues to determine whether there is an inflection point on the original curve of the re-analysis monitoring time. If there is, the inflection point on the original curve of the re-analysis monitoring time that is closest to the theoretical inflection point is selected as the target inflection point. If there is no inflection point, the comprehensive distance between the original curve of the re-analysis monitoring time and the theoretical inflection point is calculated. The most recent original point is taken as the target inflection point, and the time corresponding to the target inflection point is recorded as the target inflection point time, and the identification of the target inflection point time is stopped; the weighted sum of the normalized distance of the second derivative and the normalized distance of the position coordinate between any original point on the original curve and the theoretical inflection point is the comprehensive distance between the corresponding original point and the theoretical inflection point; the time series curve formed by the original furnace temperature at any monitoring time to be analyzed and all monitoring times under the preset second time threshold before the monitoring time to be analyzed is the original curve of the monitoring time to be analyzed, and the points on the original curve are the original points. The result of cubic spline interpolation of the original curve of the monitoring time to be analyzed is the interpolated curve of the monitoring time to be analyzed.

9. The adaptive PID temperature control method for electronic incense burners based on artificial intelligence as described in claim 7, characterized in that, The moment when the variances of both the convexity coefficient and the thermal inertia coefficient first appear to be less than the preset variance threshold is the feature convergence moment.

10. The adaptive PID temperature control method for electronic incense burners based on artificial intelligence as described in claim 7 is characterized in that, , and The expression is: ； ； ； The preset reference weight is the weight at the t-th monitoring time to be analyzed.

Images