A fire smoke temperature measurement method based on Kalman filtering and PID control

Abstract

The present invention proposes a fire smoke temperature measurement method based on Kalman filtering and PID control to solve the technical problems that the existing smoke temperature measurement methods have hysteresis and sometimes no effective measurement. The present invention uses a temperature-sensitive fire detector to measure the fire ignition area in real time to obtain fire smoke temperature data; establishes a soft measurement mathematical model based on the obtained fire smoke temperature data; uses a PID controller to process the soft measurement model to control the smoke temperature signal; and predicts the temperature of the smoke temperature signal through Kalman filtering. The present invention uses the PID control technology based on the Kalman filter to optimally estimate the smoke temperature when a fire occurs, has a good control function on the smoke temperature signal, and can make the fire smoke temperature monitoring system reach a stable state relatively quickly. The present invention can quickly and accurately measure the fire smoke temperature, shorten the prediction time of the fire, and has good accuracy and timeliness.

Description

Technical Field

[0001] The present invention relates to the technical field of fire detection, and in particular to a fire smoke temperature measurement method based on Kalman filtering and PID control, wherein the fire smoke temperature is measured based on the Kalman filtering method for high-temperature smoke generated when a fire occurs. Background Art

[0002] At present, the construction scale of high-rise buildings, traffic tunnels and integrated pipe corridors in my country is developing rapidly. Once a fire occurs in these important places, the high-temperature smoke will cause the temperature in the space to rise sharply, causing disasters. Therefore, it is very important to measure the fire detection alarm quickly and accurately.

[0003] Many scholars have conducted research on fire detection. For example, T. Windisch proposed a fire alarm that detects fires in the surrounding environment as early as possible and alarms based on fire-specific parameters such as smoke, carbon black particles, temperature or gas concentration. Cao Zhenguan proposed a fire hazard prediction method based on temperature field, using the LS-SVM (least squares support vector machine) regression algorithm to predict the temperature at the next moment to achieve the detection and prevention of fire hazards. Zhu Jingjing and others invented a traffic tunnel fire detection system that can quickly and effectively detect tunnel fires. Zhong Wei and others invented a subway station or detection and fire power prediction based on parallel temperature measurement, which measures the smoke temperature under the ceiling in real time through temperature-sensitive optical fiber to predict the size of the fire power. Yang Mingzhou proposed a multi-point sampling method for the smoke monitoring system, which extracts sample gas evenly and has a small amount of dust.

[0004] Kalman filtering plays an important role in prediction. Even if the exact nature of the model is unknown, Kalman filtering can estimate the past and current states of the signal as well as the future state. The basic idea of ​​Kalman filtering is: taking the minimum mean square error as the criterion for the best estimate, using the state space model of the signal and noise, using the estimated value of the previous moment and the observed value of the current moment to update the estimate of the state variable, and finding the estimated value of the current moment, so as to make an estimate of the signal to be processed that satisfies the minimum mean square error according to the established system equation and observation equation.

[0005] Many scholars use Kalman filtering for prediction. Zhang Zhili et al. demonstrated the application of Kalman filtering in temperature monitoring systems and established a mathematical model for the AD590 temperature sensor. The results showed that the simulation device achieved a relatively good simulation effect. Yang Bing et al. proposed a flame detection method based on the Kalman filter motion tracking algorithm, and further confirmed the fire by using features and information such as color and circularity. Lu Shengli et al. designed a multi-temperature sensor data fusion system based on Kalman filtering. The Kalman filter data fusion algorithm was used to calculate the temperature estimation value for the sampling results. The Kalman filter data fusion algorithm has high calculation accuracy and fault tolerance, can significantly improve the temperature estimation accuracy, significantly improve the credibility of the temperature measurement system, and has real-time fusion dynamic performance. Shi Raoqiao et al. proposed a NOx concentration measurement method based on Kalman filtering and data fusion technology, and verified the method using historical data from coal-fired power plants. It has a faster measurement response speed and higher measurement accuracy. Pandiv et al. proposed an algorithm that combines time series analysis and Kalman filtering to improve the prediction accuracy of wind farm wind speed and solve the problem of time series model prediction delay.

[0006] In the current research on fire smoke monitoring systems, smoke temperature measurement has problems such as hysteresis and inability to effectively measure in some places after a fire occurs. The Kalman filter method can quickly extract the required signal and has strong performance in state estimation and prediction. It has been applied in temperature monitoring, flame monitoring, temperature sensor data fusion, wind speed prediction and other fields. However, there are few studies on the application of the Kalman filter method to fire smoke temperature measurement. Summary of the invention

[0007] In view of the technical problems that the existing smoke temperature measurement methods have hysteresis and some places cannot be effectively measured after a fire occurs, the present invention proposes a fire smoke temperature measurement method based on Kalman filtering and PID control. The Kalman filtering method is combined with the fire smoke temperature measurement to solve the problems of smoke temperature measurement hysteresis in the fire smoke monitoring system and the inability to effectively measure in some places after a fire occurs, thereby achieving accurate detection and early warning.

[0008] In order to achieve the above-mentioned purpose, the technical solution of the present invention is implemented as follows: a fire smoke temperature measurement method based on Kalman filtering and PID control, characterized in that the fire ignition area is measured in real time by a temperature-sensitive fire detector to obtain fire smoke temperature data; a soft measurement mathematical model is established based on the obtained fire smoke temperature data; the soft measurement model is processed by a PID controller to control the smoke temperature signal; and the temperature of the smoke temperature signal is predicted by Kalman filtering.

[0009] The fire smoke temperature data includes the signal and noise of the fire smoke temperature and presents a discrete distribution state.

[0010] The measured fire smoke temperature signal is processed by Kalman filtering, and the fire smoke temperature signal after Kalman filtering is controlled using fire smoke temperature measurement based on Kalman filtering and PID control. The tracked fire smoke temperature signal is more stable and close to the actual fire smoke temperature result.

[0011] The method for establishing a soft measurement mathematical model is as follows: selecting the relationship between temperature data and time under a fire condition in the past, establishing a soft measurement mathematical model, using the measured control variables as the output value of the soft measurement mathematical model, and obtaining the soft measurement value of the temperature to be measured, i.e., the temperature, through mathematical calculation.

[0012] The soft sensor mathematical model is:

[0013]

[0014] And φ 0 (t) = 1,

[0015] Among them, y f (t) is the soft measurement value of temperature; φ i (t) is the input variable; k is the time delay; u j (t) is the control variable; ε(t) is the deviation, θ i are model parameters.

[0016] When no temperature measurement is available, the soft sensor mathematical model is used to monitor the temperature in real time;

[0017] When the measured value y of the temperature measurement system at time t v (t) When available, use historical data and measured values ​​y v (t), calculate the optimal estimate of the spatial state of the flue gas temperature at time td, that is, y e (td);

[0018] In temperature monitoring, the state space system of temperature is as follows:

[0019] x(t+1)=Ax(t)+w(t)

[0020] y f (t) = C f x(t)+v f (t)

[0021] y v (t) = C v x(td)+v v (td)

[0022] Where x(t) is the state vector of the flue gas temperature at time t in the state space, that is, the true value of the temperature, A is the state transfer matrix; y f (t) and y v (t) are the measured values ​​of smoke temperature based on the soft sensor mathematical model and the fire temperature measurement system, respectively, C f and C v are the state observation matrices based on the soft sensor mathematical model and the fire temperature measurement system; d is the lag of the fire temperature automatic monitoring system, w(t), v f (t) and v v (td) are process noise, measurement noise of soft measurement and measurement noise of fire temperature automatic monitoring system respectively.

[0023] The relationship between the fire smoke temperature of the fusion measurement value and the soft sensor mathematical model is:

[0024]

[0025] Among them, x f and x v It is a fire smoke temperature spatial state estimation based on the measurement of soft measurement and fire temperature automatic monitoring system using Kalman filter algorithm. e is the state estimation variance of the fused measurement, P f and P v are the state estimation variance of the soft sensor mathematical model and the fire temperature measurement system, respectively, w is the fused measurement of temperature.

[0026] The implementation method of the Kalman filter is:

[0027] For discrete domain linear systems:

[0028] x(t)=Ax(t-1)+B(u(t)+w(t))

[0029] y v (t) = Cx(t) + v(t)

[0030] Where w(t) is the fire smoke temperature process noise signal, v(t) is the fire smoke temperature measurement noise signal, A and B are state transfer matrices; y v (t) is the measured value of the flue gas temperature of the system at time t; x(t) is the state vector of the flue gas temperature at time t in the state space, that is, the true value of the temperature; x(t-1) is the true value of the flue gas temperature at time t-1 in the state space, u(t) and C are parameters, and x(t-1), w(t), and v(t) are measured by temperature sensors;

[0031] The discrete Kalman filter recursion is:

[0032] Filter gain

[0033] Forecast estimate covariance P(t|t-1)=AP(t-1)A T +BQB T

[0034] Filter error covariance P(t) = P(t|t-1)(E n -M n (t)C)

[0035] Among them, Q is a parameter, R is a parameter; E n is a unit vector;

[0036] The spatial state of the fire smoke temperature at time t, that is, the true value of the smoke temperature is:

[0037] x(t)=Ax(t-1)+M n (t)(y v (t)-CAx(t-1));

[0038] Andey e (t) = Cx(t);

[0039] In the formula, y e (t) is the optimal estimate of temperature based on Kalman filtering;

[0040] The covariance of the error is: err cov(t)=CP(t)C T .

[0041] Using the historical data of the Kalman filter algorithm and the soft measurement mathematical model, the fused measurement value is updated to obtain the optimal temperature estimate at time td, that is, the fused measurement value:

[0042] The dynamic changes of the smoke temperature space state during a fire are as follows: Among them, x f (t) is the actual value of the flue gas temperature at time t, x f (t-1) is the true value of the flue gas temperature at time t-1;

[0043] Further predictions of the fusion measurement of smoke temperature during fire are as follows: e (t|t-1)=zx e (t-1);

[0044] The Kalman filter gain is as follows:

[0045] The optimal estimate of the fire smoke temperature spatial state:

[0046] y w(t) = x f (t)+K[y w (t|t-1)-x f (t)], where K is the Kalman filter gain, y w (t|t-1) is the smoke temperature value of the fire at t-1;

[0047] The update of the fire smoke temperature space state equation is: P e =(1-K)P f .

[0048] The measured value of the no-fire temperature y v Before (t+s) is available, use the soft measurement value y at time t+1 f (t+1), soft measurement value y at time t f (t) and the fused measurement value y at time t w (t) Update the fused measurement value to obtain the fused measurement value y at time t+1 w (t+1); As the soft measurement value is updated, the fused measurement value is recursively updated until the measurement value of the fire temperature automatic monitoring system is updated at time t+s, and the calculation y is returned. e (td) loop.

[0049] The beneficial effects of the present invention are as follows: the measured fire smoke temperature signal is processed by Kalman filtering, and then the fire smoke temperature measurement method based on Kalman filtering and PID control is used to predict the fire hazard of the fire smoke temperature signal after Kalman filtering, and the PID control technology based on Kalman filter is used to optimally estimate the smoke temperature when a fire occurs, and the smoke temperature signal has a good control function, so that the fire smoke temperature monitoring system can reach a stable state relatively quickly. The present invention can quickly and accurately measure the fire smoke temperature, so that the fire smoke temperature monitoring system can reach a stable state relatively quickly, shorten the prediction time of the fire, have good accuracy and timeliness, and increase the reaction time of personnel evacuation, which is of great significance. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0051] Figure 1 It is a schematic diagram of the process of the present invention.

[0052] Figure 2 This is a schematic diagram of the implementation principle of the PID control of the present invention.

[0053] Figure 3 This is a comparison chart of the signals after Kalman filtering and smoothing in the present invention.

[0054] Figure 4 This is a comparison diagram of the PID control step response after passing through the Kalman filter of the present invention.

[0055] Figure 5 It is a schematic diagram of the corresponding relationship between the soft measurement and the temperature detector of the present invention. DETAILED DESCRIPTION

[0056] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0057] like Figure 1 As shown, a fire smoke temperature measurement method based on Kalman filtering and PID control, the steps are as follows:

[0058] Step 1: Use a temperature-sensitive fire detector to measure the fire area in real time to obtain fire smoke temperature data.

[0059] In random optimal control and random signal processing, the signal and noise of the fire smoke temperature are obtained. The signal and noise are multi-dimensional non-steady-state random processes. Due to time-varying properties, the power spectrum often changes. The Kalman filter theory uses a recursive algorithm in the time domain and uses a computer to perform data filtering processing. The fire smoke temperature data obtained in the specific implementation presents a discrete distribution state.

[0060] First, the measured fire smoke temperature signal is processed by Kalman filtering, so that the processed fire smoke temperature signal has a better filtering effect of controlling interference and measurement noise. Then, the fire smoke temperature signal after Kalman filtering is controlled by fire smoke temperature measurement based on Kalman filtering and PID control, and the tracked fire smoke temperature signal is more stable and close to the real fire smoke temperature result.

[0061] Step 2: Establish a soft sensor mathematical model based on the fire smoke temperature data obtained.

[0062] The Kalman filter has a good filtering effect on the fire smoke temperature signal, and the PID control has a good control function on the fire smoke temperature signal, which can make the measurement system reach a stable state relatively quickly. The structure principle of the control system based on Kalman filtering is as follows: Figure 2 shown.

[0063] The relationship between the temperature data and related variables (time) under the fire conditions in the past period of time is selected to establish a soft measurement mathematical model. The measured auxiliary variables (control variables) are used as the output values ​​of the model input, that is, the soft measurement values ​​of temperature.

[0064] In the measurement of soft measurement technology, the temperature measured by the temperature sensor is first obtained, and the temperature to be measured is obtained through mathematical calculation through the soft measurement mathematical model. The corresponding relationship between the soft measurement mathematical model and the input and output of the temperature detector is as follows: Figure 5 The linear external input autoregressive soft sensor mathematical model is as follows:

[0065]

[0066] And φ 0 (t) = 1,

[0067] Among them, y f (t) is the soft measurement value of temperature; φ i (t) is the input variable; k is the time delay; u j (t) is the control variable; ε(t) is the deviation, θ i are model parameters.

[0068] When there is no measurement value available from the automatic temperature monitoring system, the soft sensor mathematical model is used to monitor the temperature in real time.

[0069] When the measured value y of the temperature measurement system at time t v (t) When available, use historical data and measured values ​​y v (t), calculate the optimal estimate of the spatial state of the flue gas temperature at time td, that is, y e (td).

[0070] In temperature monitoring, the temperature can be measured directly, and the state space system of the temperature is as follows:

[0071] x(t+1)=Ax(t)+w(t)

[0072] y f (t) = C f x(t)+v f (t)

[0073] y v (t) = C v x(td)+v v (td)

[0074] In the formula, x(t) is the state vector of the flue gas temperature in the state space at time t, that is, the true value of the temperature, and A is the state transfer matrix. Due to the existence of measurement error, it can be considered that the true value of the flue gas temperature is not equal to the measured value. f (t) and y v (t) are the measured values ​​of smoke temperature based on the soft sensor mathematical model and the fire temperature automatic monitoring system, respectively, C f and C v are the state observation matrices based on the soft sensor mathematical model and the fire temperature automatic monitoring system, respectively. d is the lag of the fire temperature automatic monitoring system, w(t), v f (t) and v v (td) are process noise, measurement noise of soft measurement and measurement noise of fire temperature automatic monitoring system respectively.

[0075] The mathematical model of soft measurement and the automatic fire temperature monitoring system are relatively independent. It can be considered that the fire smoke temperature measurement and soft measurement are unrelated. The relationship between the fire smoke temperature of the fusion measurement, soft measurement and automatic fire temperature monitoring system is expressed as follows:

[0076]

[0077] Among them, x f and x v It is a fire smoke temperature spatial state estimation based on the measurement of soft measurement and fire temperature automatic monitoring system using Kalman filter algorithm. e is the state estimation variance of the fused measurement, P f and P v are the state estimation variance of the soft sensor mathematical model and the temperature measurement system, respectively, w is the fusion measurement value of temperature, and it can be considered that y w It is the optimal estimation of temperature space by combining soft measurement and fire temperature automatic monitoring system.

[0078] Step 3: Use the PID controller to process the soft measurement mathematical model, run the calculation through the computer, and control the flue gas temperature signal.

[0079] The PID control response based on Kalman filtering is calculated through computer operation, the system response is faster, the system can converge to a steady state better, and the control effect is better.

[0080] Step 4: Predict the temperature of the flue gas temperature signal through Kalman filtering.

[0081] The Kalman filter is an optimized autoregressive data processing algorithm. When solving the problem of fire smoke temperature measurement, an accurate soft-sensing mathematical model is established, and then the Kalman filter is designed and implemented based on the soft-sensing mathematical model.

[0082] For discrete domain linear systems:

[0083] x(t)=Ax(t-1)+B(u(t)+w(t))

[0084] y v (t) = Cx(t) + v(t)

[0085] There is a deviation between the smoke temperature value measured by the temperature measurement system composed of heat-sensing fire detectors and the actual smoke temperature value. These deviations are regarded as noise signals. In the formula, w(t) is the fire smoke temperature process noise signal, v(t) is the fire smoke temperature measurement noise signal, and A and B are state transfer matrices. v (t) is the measured value of the flue gas temperature of the system at time t. x(t) is the state vector of the flue gas temperature at time t in the state space, that is, the true value of the temperature. x(t-1) is the true value of the flue gas temperature at time t-1 in the state space. u(t) and C are parameters. x(t-1), w(t), and v(t) are measured by temperature sensors.

[0086] The discrete Kalman filter recursion is as follows:

[0087] Filter gain

[0088] Forecast estimate covariance P(t|t-1)=AP(t-1)A T +BQB T

[0089] Filter error covariance P(t) = P(t|t-1)(E n -M n (t)C)

[0090] Among them, Q is a parameter, R is a parameter, both of which are taken as 1; E n is a unit vector.

[0091] The spatial state of the fire smoke temperature at time t, that is, the true value of the smoke temperature is

[0092] x(t)=Ax(t-1)+M n (t)(y v (t)-CAx(t-1));

[0093] Andey e (t) = Cx(t)

[0094] In the formula, y e (t) is the optimal estimate of temperature based on Kalman filtering.

[0095] The covariance of the error is: err cov(t)=CP(t)C T .

[0096] Kalman filter can directly process the smoke temperature signal, which requires the customer to overcome the barriers of computer operation calculation. Kalman filter has a good filtering effect on control interference and measurement noise. The fire smoke temperature control interference signal w(t) and the measurement noise signal v(t) are selected as white noise signals with an amplitude of 0.01, and the input signal amplitude is 1.0 and the frequency is a sine signal of 1.5HZ.

[0097] The fire smoke temperature is set to 500°C, and the signal results obtained by measuring the smoke temperature are tracked: assuming that the measured value of the fire smoke temperature at the t-1th moment is 498°C, and the actual temperature of the fire smoke temperature is 500°C, the deviation of the measured value is 2°C, that is, the covariance is 4. The smoke temperature value at the t-1th moment is used to predict the smoke temperature at the tth moment, and its expected deviation is obtained, and then the Kalman gain is calculated. Using the observed value of the smoke temperature at the tth moment, the estimated value of the smoke temperature is obtained through numerical calculation, and the Kalman estimated value is closer to the true value. At this point, the system automatically updates the deviation at the tth moment. Finally, based on the estimated value of the smoke temperature and the deviation at the tth moment, the observed data at the next moment can continue to be updated and processed. In this way, the Kalman filter continuously recursively estimates the optimal smoke temperature value and obtains Figure 3 Results shown for the Kalman filtered signal and the smoothed signal. Figure 3 It can be seen that when the fire smoke temperature signal is filtered by Kalman filtering, the smaller the Kalman gain, the more reliable the fire smoke temperature prediction value, and the closer the optimization value is to the fire smoke temperature prediction value. The larger the Kalman gain, the more reliable the fire smoke temperature measurement value, and the closer the optimization value is to the smoke temperature measurement value.

[0098] When controlling the smoke temperature in a fire, there are a lot of pseudo-random signals in the temperature measurement due to the uncertainty of the smoke temperature. Using Kalman filtering to track the temperature measurement can get closer to the target value. The fire smoke temperature signal obtained by tracking with Kalman filtering is more stable and close to the actual fire smoke temperature result.

[0099] The PID control operation of the Kalman filter is adopted, the sampling time is 1ms, the interference signal w(t) and the measurement noise signal v(t) of the fire smoke temperature are controlled to be white noise signals with an amplitude of 0.002, and the input signal amplitude is a step signal. The Kalman filter is used to filter the fire smoke temperature measurement signal. Take Q=1, R=1, and the simulation time is 1s. The results are as follows Figure 4 As shown. Figure 4 It can be obtained that the Kalman filter is used to filter the fire smoke temperature measurement signal.

[0100] The Kalman filter algorithm is used to utilize the historical data of the soft measurement mathematical model to realize the update of the fused measurement value and obtain the optimal estimate of the temperature at time td, that is, the fused measurement value.

[0101] The dynamic changes of the smoke temperature space state during a fire are as follows:

[0102] Among them, x f (t) is the actual value of the flue gas temperature at time t, x f (t-1) is the true value of the flue gas temperature at time t-1.

[0103] Further predictions of the fusion measurement of smoke temperature during fire are as follows: e (t|t-1)=zx e (t-1).

[0104] The Kalman filter gain is as follows:

[0105] The optimal estimate of the fire smoke temperature spatial state is as follows:

[0106] y w (t) = x f (t)+K[y w (t|t-1)-x f (t)], where K is the Kalman filter gain, y w (t|t-1) is the smoke temperature value of the fire at t-1.

[0107] The update of the fire smoke temperature space state equation is as follows: e =(1-K)P f .

[0108] In the absence of fire, the measured value of the temperature automatic monitoring system is y v Before (t+s) is available, use the soft measurement value y at time t+1 f (t+1), soft measurement value y at time t f (t) and the fused measurement value y at time t w (t) Update the fused measurement value to obtain the fused measurement value y at time t+1 w (t+1). As the soft measurement value is updated, the fused measurement value is recursively updated until the measurement value of the fire temperature automatic monitoring system is updated at time t+s, and the calculation y is returned. e (td) loop.

[0109] Figure 1The output temperature signal is used to judge the fire hazard, predict whether there is a fire trend, and realize it by comparing and running the calculation through a computer program. The present invention performs Kalman filtering on the measured fire smoke temperature signal, so that the manufactured fire smoke temperature signal has a better filtering effect of controlling interference and measurement noise. The present invention uses the Kalman filter-based PID control technology to control the fire smoke temperature signal after Kalman filtering, and the tracked fire smoke temperature signal is more stable and close to the actual fire smoke temperature result. The present invention is based on the PID control response of Kalman filtering, the system response is faster, and the system can better converge to a steady state, and the control effect is better.

[0110] The PID fire smoke temperature control based on Kalman filtering of the present invention has a fast system response, can better make the system converge to a steady state, and has a better control effect. The PID step response with Kalman filtering effectively suppresses disturbances and noise, the response curve basically has no overshoot, the adjustment time is within 0.2s, and has good adaptability and anti-interference when the control parameters change. The present invention solves the technical problems of inaccuracy or long time lag in the existing fire temperature measurement method, and achieves the technical effect of improving the accuracy of fire smoke temperature measurement and simple calculation.

[0111] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, improvements, etc. made within the spirit and principle of the present invention should be included in the protection scope of the present invention.

Claims

1. A fire smoke temperature measurement method based on Kalman filtering and PID control, characterized in that: The fire area is measured in real time by the temperature-sensing fire detector to obtain the fire smoke temperature data; a soft measurement mathematical model is established based on the obtained fire smoke temperature data; the soft measurement model is processed by the PID controller to control the smoke temperature signal; the temperature of the smoke temperature signal is predicted by the Kalman filter; The method for establishing the soft measurement mathematical model is as follows: selecting the relationship between temperature data and time under a fire condition in a certain period of time in the past, establishing the soft measurement mathematical model, using the measured control variable as the output value of the soft measurement mathematical model, and obtaining the soft measurement value of the temperature to be measured, i.e., the temperature, through mathematical calculation; The soft sensor mathematical model is: And φ0(t)=1, Among them, y f (t) is the soft measurement value of temperature; φ i (t) is the input variable; k is the time delay; u j (t) is the control variable; ε(t) is the deviation, θ i are model parameters.

2. The fire smoke temperature measurement method based on Kalman filtering and PID control according to claim 1 is characterized in that: The fire smoke temperature data includes the signal and noise of the fire smoke temperature and presents a discrete distribution state.

3. The fire smoke temperature measurement method based on Kalman filtering and PID control according to claim 2 is characterized in that: The measured fire smoke temperature signal is processed by Kalman filtering, and the fire smoke temperature signal after Kalman filtering is controlled using fire smoke temperature measurement based on Kalman filtering and PID control. The tracked fire smoke temperature signal is more stable and close to the actual fire smoke temperature result.

4. The fire smoke temperature measurement method based on Kalman filtering and PID control according to any one of claims 1 to 3, characterized in that: When no temperature measurement is available, the soft sensor mathematical model is used to monitor the temperature in real time; When the measured value y of the temperature measurement system at time t v (t) When available, use historical data and measured values ​​y v (t), calculate the optimal estimate of the spatial state of the flue gas temperature at time td, that is, y e (td); In temperature monitoring, the state space system of temperature is as follows: x(t+1)=Ax(t)+w(t) y f (t)=C f x(t)+v f (t) y v (t)=C v x(t-d)+v v (t-d) Where x(t) is the state vector of the flue gas temperature at time t in the state space, that is, the true value of the temperature, A is the state transfer matrix; y f (t) and y v (t) are the measured values ​​of smoke temperature based on the soft sensor mathematical model and the fire temperature measurement system, respectively, C f and C v They are based on the soft sensor mathematical model and the state observation matrix of the fire temperature measurement system; d is the hysteresis of the fire temperature automatic monitoring system, w(t), v f (t) and v v (td) are process noise, measurement noise of soft measurement and measurement noise of fire temperature automatic monitoring system respectively.

5. The fire smoke temperature measurement method based on Kalman filtering and PID control according to claim 4 is characterized in that: The relationship between the fire smoke temperature of the fusion measurement value and the soft sensor mathematical model is: Among them, xf and x v It is a fire smoke temperature spatial state estimation obtained by using the Kalman filter algorithm based on the measurement of soft measurement and fire temperature automatic monitoring system. e is the state estimation variance of the fused measurement, P f and P v are the state estimation variance of the soft sensor mathematical model and the fire temperature measurement system, respectively, w is the fused measurement of temperature.

6. The fire smoke temperature measurement method based on Kalman filtering and PID control according to any one of claims 1 to 3 and 5, characterized in that: The implementation method of the Kalman filter is: For discrete domain linear systems: x(t)=Ax(t-1)+B(u(t)+w(t)) y v (t)=Cx(t)+v(t) Where w(t) is the fire smoke temperature process noise signal, v(t) is the fire smoke temperature measurement noise signal, A and B are state transfer matrices; y v (t) is the measured value of the flue gas temperature of the system at time t; x(t) is the state vector of the flue gas temperature at time t in the state space, that is, the true value of the temperature; x(t-1) is the true value of the flue gas temperature at time t-1 in the state space, u(t) and C are parameters, and x(t-1), w(t), and v(t) are measured by temperature sensors; The discrete Kalman filter recursion is: Filter gain Forecast estimate covariance P(t|t-1)=AP(t-1)A T +BQB T Filter error covariance P(t) = P(t|t-1)(E n -M n (t)C) Among them, Q is a parameter, R is a parameter; E n is a unit vector; The spatial state of the fire smoke temperature at time t, that is, the true value of the smoke temperature is: x(t)=Ax(t-1)+M n (t)(y v (t)-CAx(t-1)); Andey e (t) = Cx(t); In the formula, y e (t) is the optimal estimate of temperature based on Kalman filtering; The covariance of the error is: err cov(t)=CP(t)C T .

7. The fire smoke temperature measurement method based on Kalman filtering and PID control according to claim 6 is characterized in that: Using the historical data of the Kalman filter algorithm and the soft measurement mathematical model, the fused measurement value is updated to obtain the optimal temperature estimate at time td, that is, the fused measurement value: The dynamic changes of the smoke temperature space state during a fire are as follows: Among them, x f (t) is the actual value of the flue gas temperature at time t, x f (t-1) is the true value of the flue gas temperature at time t-1; Further predictions of the fusion measurement of smoke temperature during fire are as follows: e (t|t-1)=zx e (t-1); The Kalman filter gain is as follows: The optimal estimate of the fire smoke temperature spatial state: y w (t) = x f (t)+K[y w (t|t-1)-x f (t)], where K is the Kalman filter gain, y w (t|t-1) is the smoke temperature value of the fire at t-1; The update of the fire smoke temperature space state equation is: P e =(1-K)P f .

8. The fire smoke temperature measurement method based on Kalman filtering and PID control according to claim 7 is characterized in that: The measured value of the no-fire temperature y v Before (t+s) is available, use the soft measurement value y at time t+1 f (t+1), soft measurement value y at time t f (t) and the fused measurement value y at time t w (t) Update the fused measurement value to obtain the fused measurement value y at time t+1 w (t+1); As the soft measurement value is updated, the fused measurement value is recursively updated until the measurement value of the fire temperature automatic monitoring system is updated at time t+s, and the calculation y is returned. e (td) loop.

Images