Research on Modeling and Deep Peak Regulation Control of a Combined Heat and Power Unit

This paper firstly proposes a nonlinear dynamic model of a combined heat and power (CHP) unit with absorption heat pump (AHP) and bypass systems, the unknown parameters are determined based on design data and perturbation test. Simulation results show that the model can reveal the couplings of AHP and bypass systems to the CHP unit, and provide model support for controller design. On the basis of the model, this paper further proposes a deep peak regulation control strategy, in which, the generalized predictive control algorithm with feedforward-feedback structure is adopted to fundamentally solve the control problems of large delay and inertia on the boiler side, and overcome known disturbances on the turbine side; the ratio of the first stage pressure to the exhaust pressure from high pressure cylinder is adopted to control the high pressure bypass. The deep peak regulation process is divided into two stages: 1) CHP and AHP are used for heating when their heating capacity can meet the heat load requirements of residents, 2) as the unit load is further reduced, bypass mode is activated for heating when the steam flow entering the low pressure cylinder is lower than its cooling flow. Simulation results show that the strategy can meet the heat load requirements of residents and ensure the safe and stable operation of the turbine when the unit is in deep peak regulation condition.


I. INTRODUCTION
In recent years, with the support of policies [1], [2], the installed capacity of wind power has been growing rapidly in China [3], however, the phenomenon of abandoning wind power is serious [4], especially in northeast China [5]. The main reason lies in the scarcity of power supply that can participate in deep peaking, such as hydropower and condensing power [6]. Instead, there are many combined heat and power (CHP) units in these area, but these units need to balance the heat load requirements of residents. In this case, CHP units are prone to excessive power generation when the electrical load demand is low, thus making the space for wind power integration insufficient. Obviously, if the ''ordering power by heat'' constraint of CHP units is decoupled during this periods [7], the power generated by CHP units can be reduced, in this case, huge grid-connected space can be created for wind power. Consequently, decoupling the constraint of CHP which provides a theoretical basis for the design of CHP units with AHP. Reference [19], [20] analyzed the heating performance of an AHP, clarified the change and distribution of energy during heating, and proposed a new CHP system. Results show that the new system can improve the efficiency and reduce the energy consumption. In terms of scheduling, [21] proposed a scheduling model for a CHP unit based on heat transfer principle. This model can adapt to the situation that the CHP unit deviates from the design condition. Reference [22] described a tractable integrated heat and electricity dispatch model, this model focuses on the thermal inertia characteristics of pipelines and buildings to increase flexibility.
In summary, the research on modeling and deep peak regulation control of CHP units is insufficient, especially for CHP units equipped with AHP and bypass systems. Therefore, this paper firstly proposes a nonlinear dynamic model of a CHP unit with AHP and bypass systems, gives the method for determining the unknown parameters, and verifies the effectiveness of the model. Based on the model, a deep peak regulation control is proposed based on generalized predictive control algorithm. Simulation results show that the strategy can achieve deep peak regulation of the CHP unit.
The paper is organized as follows. Section 2 introduces the working principle of a CHP unit with AHP and bypass systems, proposes a nonlinear dynamic model of the CHP unit, designs and verifies a deep peak regulation control strategy for the CHP unit. Section 3 elaborates conclusions.

II. MODELING, CONTROL AND ANALYSIS A. WORKING PPRINCIPLE
As shown in FIGURE 1, in addition to the extraction heating mode, the CHP unit has also been improved to some extent. Firstly, a bypass heating system is added. Part of main steam is desuperheated and decompressed by high pressure bypass (HPB), and then mixed with the exhaust steam from the high pressure cylinder (HPC), the mixed steam is sent to the reheater for reheating. After reheating, part of reheat steam is further desuperheated and decompressed by low pressure bypass (LPB) and mixed with the extraction steam from the intermediate pressure cylinder (IPC), this mixed steam is actually prepared for heating. Part of the heating steam is used as the driving steam to enter the AHP, while the other part of the heating steam directly enter the heater for heating. Secondly, an AHP heating system is added. Driven by the steam, the heat consists in circulating water can be recycled by AHP heating system, the recycled heat is then used to heat the return water initially. After the initial heating, the return water is divided into two parts: one part enters the heater to continue absorb heat, thereby further increasing the temperature of return water; the other part directly enters the header for mixing, so that the temperature of the mixed water can meet the heat load requirements of residents.
For bypass system [10], the steam flows into the cold section of the reheater increases with the opening of the HPB valve, since the resistance of the reheater remains unchanged, more steam flows through the reheater, resulting in the differential pressure of the reheater to increase. If the LPB valve is not opened in time, the exhaust steam pressure from HPC will be increased rapidly. Since the temperature and pressure of the steam are coupled, the exhaust steam temperature from HPC will also be increased. Excessive pressure and temperature of exhaust steam will cause changes in the flow characteristics of the HPC, as a result, the balance of the original axial thrust will be broken. In this case, the strength of the final stage blades for the HPC will also be affected. Therefore, it is necessary to open the LPB valve timely to ensure that the exhaust parameters from the HPC is within a reasonable range.
For AHP systems [23], it is necessary to raise the back pressure of CHP units to 7kPa to ensure that the temperature of circulating water can meet the design parameters of AHP systems. Since the increase of the back pressure directly increases the cooling flow of the LPC, the heating capacity of the CHP units transformed by AHP systems will be reduced, especially under low load conditions, the situation that the AHP heating mode and the extraction heating mode cannot be put into operation at the same time may occur. That is: the extraction heating mode may not be able to put into operation when the AHP has been put into operation, which is due to the limitation of the cooling flow for the LPC; the same reason, the back pressure of CHP units may not be able to raise when extraction heating mode is put into operation, as a result, the temperature of circulating water will be insufficient for normal operation of AHP systems.

B. DYNAMIC MODEL 1) MODEL STRUCTURE
Currently, the boiler model is relatively mature [24], [25], but the model with bypass systems has not yet appeared. Therefore, this paper firstly improves the boiler model, the improved model includes a pulverizing system model (Eq.(1)), a drum model (Eq.(2)), and a main steam pipe model (Eq.(3)) and a reheat steam pipe model (Eq.(4) and Eq. (5)).
where q b is coal feed flow, t/h; q f is the amount of coal entering the boiler, t/h; τ is the delay time of the coal feeder, s; T f is the inertia time of the coal mill and the separator, s; C b is the storage coefficient of steam drum, (t/h·s)/MPa. p b is the drum pressure, MPa; p t is main steam pressure, MPa; C t is the storage coefficient of main steam pipe, (t/h·s)/MPa. u t is the position of main steam valve, %; u HPB i is the positon of HPB valve, %; C r is the storage coefficient of reheat steam pipe, (t/h·s)/MPa; D r is the reheat steam flow, t/h; p r is the reheat steam pressure, MPa; u LPB i is the positon of LPB valve, %; τ , T f , C b , C t , C r are dynamic parameters that need to be determined, K 1 , . . . , K 8 are static parameters that need to be determined.
The ratio of the first stage pressure to the exhaust pressure from HPC is a parameter that needs to be monitored when bypass heating mode is put into operation (Eq.(6)).

R HPC
The saturation temperature of water in the heater reflects the balance between a) the heat consists in exhaust steam from IPC, the heat consists in the steam generated by the LPB and b) the heat entering the LPC, the heat entering the heater. Thus the saturation temperature of water in the heater is selected as a state variable. Thus, the model of heater is as show in Eq. (7) and Eq. (8).  (8) where M HTR sw is the temperature-based storage coefficient of the heater, MJ/ • C; θ HTR sw is the saturation temperature of water in the heater, • C; Q HTR sw is the heat released by the steam entering the heater, MJ/s; q HTR sw is the return water flow entering the heater, t/h; c p,x is the specific heat capacity of the return water, MJ/(t· • C); θ HTR is the terminal difference of the heater, • C; θ HTR o is the outlet water temperature of the heater, • C; θ HTR i is the inlet water temperature of the heater, • C.
Since there is a large inertia and delay in the transmission process of θ HTR sw , θ HTR sw is not suitable to be taken as a state variable. For this reason, [10]  The heat released by the steam entering the heater Q HTR h is actually related to the enthalpy of exhaust steam from the IPC, the saturation enthalpy of water in the heater, and the steam flow entering the heater. Therefore, the heat released by the steam entering the heater Q HTR h can be calculated as: where The coefficient of performance (COP) is a key parameter reflecting the performance of AHP, which can be defined as follows: where Q 1 is the heat output from AHP, MW; Q 2 is the heat absorbed from circulating water, MW; W is the heat released by driving steam, MW; Q 2 and W can be calculated according to Eq. (17) and (18) respectively.
where q AHP c is the circulating water flow, t/h; θ AHP ci is the inlet temperature of circulating water, • C; θ AHP co is the outlet temperature of circulating water, • C; D AHP h is the driving steam flow, t/h; h AHP sw is the saturation enthalpy of water in AHP, kJ/kg. Based on the design parameters of the AHP (TABLE 2), Q 2 , W and COP can be calculated: Q 2 = 21.72MW, W = 32.035MW, COP = 1.68.
Assume that the COP of the AHP is unchanged during operation, then the heat transfer process of AHP can be described as show in Eq. (19), in this case, the outlet temperature of heating water can be obtained (Eq. (20)). (20) where θ AHP hi is the inlet temperature of heating water, • C; θ AHP ho is the outlet temperature of heating water, • C. In actual application, the heating water output from AHP is divided into two parts, one part of the water enter the heater for further heating, while the other part of the water enters the heating water header. These two parts of water are mixed in the header. Ignoring the change in the specific heat capacity of the water, the temperature of the mixed water can be calculated as show in Eq. (21).
where, θ mix is the temperature of mixed water, • C; q HTR h is the water flow further heated by the heater, t/h, considering that the total amount of heating water of the CHP unit is 9000t/h, then the water flow into the header is 9000 − q HTR h , t/h; θ HTR o is the outlet water temperature of the heater, • C; θ AHP ho is the outlet temperature of heating water from the AHP, • C.
The power generated by each pressure cylinder are basically same as their internal power. Ignoring the heat loss of each pressure cylinder, the internal power of each pressure cylinder can be calculated according to Eq. (22). is the kth extraction steam enthalpy of the pressure cylinder, kJ/kg; n is the number of extraction of each pressure cylinder. All parameters in Eq. (22) can be found in the thermal equilibrium diagram (TABLE 3, TABLE 4 and TABLE 5).
Based on the calculation results, the relationship between internal power of each pressure cylinder and their inlet steam flow can be fitted as show in Eq. (23)- (25), and the total power output of the CHP units can be calculated as show in Eq. (26).
; N e is the total power output of the unit, MW; T t is the inertia time of turbine, and is a unknown parameter that need to be determined.
In summary, the proposed model is as follows: where h ; the output of model are p t , N e and θ mix ; the unknown parameters that need to be determined are K 1 , . . . ,

2) PARAMETER DETERMINATION
On the basis of the design parameters of the unit (TABLE 6), the unknown static parameters K 1 to K 12 can be calculated as show in Eq.(28) -(40), where the subscript RG represents rated generation condition and RH represents rated heating condition.
VOLUME 8, 2020 The dynamic unknown parameters of the model can to be obtained by perturbation test. Considering that the parameter T f , τ , C h , T t of the same type of unit have been studied in [26], this paper directly refers to them, that is T f = 120s, τ = 15s, C h = 160MJ/MPa, T t = 12s. The storage coefficient of drum and main steam pipe are generally determined by perturbation test of main steam valve. During the test, the drum pressure and the main steam flow are recorded, and the total storage coefficient can be calculated as follows: The total storage coefficient obtained by the testC b = 5308(t/h·s)/MPa. The storage coefficient of the drum accounts for 90% of the total storage coefficient, while that of main and reheat steam pipe accounts for 10% of the total storage coefficient. Assume that the storage coefficients of the main steam pipe and the reheat steam pipe account for 5% respectively, then the storage coefficient of the drum C b = 4777.2(t/h·s)/MPa, that of main steam pipe C t = 265.4(t/h·s)/MPa, and that of reheat steam pipe C r = 265.4(t/h·s)/MPa. In summary, the model parameters obtained in this paper are shown in TABLE 1.

3) MODEL VERIFICATION
The parameters of the initial conditions are: the coal feed flow is 217.25t/h; the position of main steam valve is 83.16%; the inlet valve position of LPC is 32.4%; the total power output of the unit is 256.29MW; the main steam pressure is 16.7MPa; the exhaust pressure from IPC is 0.49MPa; the extraction steam flow for heating is 500t/h, of which, 397.265t/h of steam enters the heater, 102.735t/h of steam enters the AHP systems; the temperature of return water is 40 • C; the total flow of return water is 9000t/h, 2859.50t/h of the water enters the heater for further heating, 6140.50t/h of water enters the header for mixing.  are changed. It can be seen from the figure that: 1) since the response rate of the boiler is slow, there is different degrees of inertia and delay in response of p t , N e and θ mix when q b decreases by 10%, especially for that of θ mix ; 2) the p t drops rapidly and stabilizes at the new state when u t increases by 10%, the N e and θ mix return to their initial state after a small overshoot, which is due to the shorttime release of the boiler energy storage; 3) since the opening of u HPB i reduces the amount of steam entering the HPC, the power output of the unit N e is reduced. The reheat steam pressure is increased simultaneously during the opening of u HPB i , further increasing the saturation temperature of exhaust steam from IPC, assume that the erminal difference of the heater is constant, then the outlet of water temperature from the heater is increased, eventually leading to an increase in the temperature of the mixed water.  entering the heater is increased, thereby increasing the outlet water temperature from the heater; 2) since the opening of the u LPC i increases the amount of steam entering the LPC, the power output of the unit is increased, but at the same time, the amount of steam entering the heater is decreased, thereby decreasing the outlet water temperature from the heater; 3) since the decrease of u AHP i increases the exhaust pressure from the IPC, thereby allowing more steam to enter the LPC and the heater, as a result, the power output of the unit and the temperature of the mixed water are increased.  are changed. It can be seen from the figure that: 1) the exhaust pressure from the IPC is decreased when θ AHP hi is decreased, thereby making the amount of steam entering the LPC decrease, as a result, the power output of the unit is decreased; since the heat absorbed from the steam is substantially constant, the decrease of θ AHP hi decreases θ mix ; 2) the decrease of q HTR h decreases the extraction steam pressure from the IPC, thereby making the amount of steam entering the LPC increase C b , as a result, the power output of the unit is increased; since the heat absorbed from the steam is decreased, the temperature of mixed water is decreased.  flow of the HPC, but increases the inlet steam flow of the IPC, considering that the axial thrust of cylinders is generally proportional to their steam intake, the balance of the original axial thrust will be broken. Besides, the increase of u HPB i decreases the R HPC , which will make the exhaust steam pressure and temperature from the HPC increase, in this case, the strength of the final stage blades of the HPC will be weakened. 2) the increase of u LPB i decreases the inlet steam flow of the IPC, in the same way, the balance of the original axial thrust will also be broken. Besides, the increase of u LPB i increases the R HPC , which will make the exhaust steam pressure and temperature from the HPC decrease, in this case, the strength of the final stage blades of the HPC will also be weakened. Therefore, it is necessary to coordinate the u HPB i and u LPB i to ensure the safe and stable operating of the CHP unit.

C. CONTROL STRATEGY
The design and transformation of the CHP unit's main and auxiliary equipment are the basis for improving its flexibility, and deep peak regulation control is the key. Considering there is large delay and inertia in boiler pulverization and combustion, the traditional PID control algorithm cannot fundamentally solve the control problem of such system. Therefore, this paper takes the predictive control algorithm as the core to design the deep peak regulation control strategy. In order to overcome the effect of known disturbances on the control system, the generalized predictive control algorithm with feedforward-feedback structure is adopted in this paper. The control law is solved in a stair-like manner to avoid the irreversibility of the matrix and ensure the real-time performance of the algorithm. The specific control algorithm and strategy are as follows:

1) PREDICTIVE CONTROL
Consider the following controlled autoregressive integrated moving average model [27], [28]: VOLUME 8, 2020 where, A q −1 , B q −1 , C q −1 and D q −1 are n, n b , n c and n d order polynomials of q −1 , respectively; y (t) is the output of the system; u (t) is the input of the system; v (t) is the feedforward of the system for decoupling; ξ (t) is the white noise; is the difference operator, = 1 − q −1 . Introducing a Diophantine equation: where, E j q −1 = e j,0 + e j,1 q −1 + · · · + e j,j−1 q −(j−1) , F j q −1 = f j,0 + f j,1 q −1 + · · · + f j,n q −n . After solving the Diophantine equation, the prediction model can be written as: where, Adopt a stair-like manner so that and the prediction model in Eq. (44) can be written as: where,G = G 1, β, · · · , β N u −1 T . Assume that the objective function: where, w is the softening set point of the system; N 1 is the starting value of the prediction horizon; N 2 is the end value of the prediction horizon; N u is the control horizon; λ is the control weight.
Based on the prediction model in Eq. (45), the objective function can be written as: Minimize the function ∂J ∂δ = 0, the control law can be obtained as:

2) STRATEGY DESIGN
Based on the generalized predictive control algorithm, a deep peak regulation control strategy is designed as shown in FIGURE 6. This strategy includes a GPC controller for main steam pressure, five PID controllers for unit load, high pressure bypass, low pressure bypass, heating extraction, absorption heat pump respectively, the strategy of each controller is as: Main steam pressure GPC controller: 1) the static feedforward from N e(sp) is a linear function that converts N e(sp) into coal feed flow roughly, which is used to quantify coal feed flow during load changing process; 2) the dynamic feedforward from AGC and N e(sp) is mainly used to pre-regulate coal feed flow, thus overcoming the problem that the feedback adjustment of the controller is slow in the initial stage of load changing process; 3) the known disturbance feedforward from u t and u HPC i is mainly used for the decoupling control of the unit, so as to facilitate the utilization and supplementation of the energy storage timely. The main steam pressure is a signal that can represent the energy balance between the boiler and the turbine, therefore, p t is selected as the controlled variable.
Unit load PID controller: this controller is generally independent in the DEH system and is mainly used to track the grid load command. Therefore, N e is selected as the controlled variable.
High pressure bypass PID controller: the exhaust steam pressure from HPB should track the exhaust steam pressure from HPC when the HPB is put into operation, but these two pressures are coupled, the increase of the exhaust steam pressure from HPB will increase the exhaust steam pressure from HPC, in this case, positive feedback control will occur and the safe operation of the turbine cannot be guaranteed. The ratio of the first stage pressure to the exhaust pressure from HPC is a constant during load changing process, if this ratio is kept within a certain range, the safe operation of the turbine can be guaranteed. Therefore, R HPC is selected as the controlled variable.
Low pressure bypass PID controller: the extraction steam flow cannot meet the heat load requirements of residents when the unit is in the deep peaking condition, in this case, the low pressure bypass can be opened to compensate for heating. Therefore, θ mix is selected as the controlled variable. It should be noted that the heat pump can only be put into operation normally if the back pressure of the unit is increased to 7kPa. In this case, the minimum cooling flow of the LPC is 180t/h. Since the cooling flow is the key to ensure the safe operation of the LPC, this paper assumes that the depth peaking signal (Flag) is triggered when D LPC i(pv) < 200t/h. Heating extraction PID controller: the heating capacity of extraction steam can meet the heat load requirements of residents when the unit is in the high load condition, in this case, the steam flow entering the LPC is generally higher than its cooling flow. However, the heating capacity of extraction steam may not meet the heat load requirements of residents when the CHP unit enters the low load condition, in this case, the steam flow entering the LPC may be lower than the cooling flow. Based on the analysis, the heating extraction PID controller actually includes two control loops, one of which controls the heating water temperature (θ mix ) in the high load condition, and the other loop controls the steam flow entering the LPC (D LPC i ), so as to ensure the safe and stable operation of the LPC.
Absorption heat pump PID controller: this controller is relatively simple, as long as it operates under the design parameters, its heating efficiency can be guaranteed. Therefore, θ AHP hi is selected as the controlled variable.

3) CONTROL VERIFICATION
During the simulation, the sampling time is 1s, the ramp rate is 6.6MW/min (2% Pe/min). The parameters of the main steam pressure GPC controller are: N 1 = 70, N 2 = 100, N u = 10, λ = 0.01, α = 0.98, β = 70; the parameters of the unit load PID controller are: K P = 5, K I = 0.01; the parameters of the high pressure bypass PID controller are: K P = 5, K I = 0.4; the parameters of the low pressure bypass PID controller are: K P = 6, K I = 0.2; the heating extraction PID controller includes two control loops, the parameters of control loop for θ mix are: K P = 0.1, K I = 0.05, the parameters of control loop for D LPC i are: K P = 0.5, K I = 0.01; the parameters of absorption heat pump PID controllers are: K P = 8, K I = 0.04. The simulation results are shown in FIGURE 7, FIGURE 8, FIGURE 9, FIGURE 10 and FIGURE 11. FIGURE 7 shows the curve for each controlled variable. It can be seen from the figure that: each controlled variable can closely track its set point, which shows that the control system can adapt to changes in conditions. The mixed water temperature can still be maintained at the design value of 91 • C when the CHP unit enters the low load condition (180MW) (FIGURE 7 (c)). Assuming that the minimum load output of the CHP unit without AHP and bypass systems is 230MW, then this deep peak regulation control can free up 50MW of grid space for wind power.     (FIGURE 8 (a)). Since the VOLUME 8, 2020  sliding pressure operation mode is adopted in the simulation, the steam flow entering the turbine can decrease naturally with the decrease of main steam pressure, therefore, the main steam valve is maintained at a higher opening degree (FIGURE 8 (b)). Since the heating capacity of extraction and AHP can meet the heat load requirements of residents when the unit is in the high load condition, the opening of u LPC i is reduced during the first load reduction process. However, as the unit load is further reduced, the heating capacity of the extraction and AHP cannot meet the heat load requirements, in other words, the steam flow entering the LPC has been below the cooling flow, in this case, u LPC i needs to ensure that the steam flow entering the LPC is higher than its cooling flow, therefore, the opening of u LPC i is not further reduced (FIGURE 8 (c)). FIGURE 9 shows the tracking and bumpless switching of heating extraction controller. The peaking signal (Flag) is triggered when the steam flow entering the LPC is lower than 200t/h (FIGURE 9(a)), and then the control loop for θ mix is switched to the control loop for D LPC i (FIGURE 9(b) and (c)). The cooling flow can always be kept above 180t/h, therefore, the control system can ensure the safe and stable operation of the LPC (FIGURE 9(c)).
FIGURE 10 shows the tracking and bumpless switching of high and low pressure bypass controllers. As the unit load is further reduced, the heating capacity of the extraction and AHP cannot meet the heat load requirements. In this case, the peaking signal Flag is triggered (FIGURE 10(a)), the bypass system is put into operation. In the bypass heating mode, the low pressure bypass controller is switched to control θ mix , and the high pressure bypass controller is switched to control R HPC . There is a certain range of fluctuations in θ mix , but it can meet the requirement of engineering applications (FIGURE 10(c)). Compared with the curve of u LPB i , that of u HPB i is smoother and more stable, thereby, the effect of HPB on main steam pressure is weakened. FIGURE 11 shows the curves of the controlled and control variables of the AHP. It can be seen from the figure that the   outlet temperature of the AHP can track its set point well, especially when the peaking signal Flag is triggered.

III. CONCLUSIONS
The operation mode of CHP units based on ''ordering power by heat'' is the main reason leading to severe wind power abandonment in northeast China. Decoupling the constraint of ''ordering power by heat'' for CHP units is an important way to solve this problem. AHP and bypass heating methods are two important ways to decouple this constraint. Firstly, a nonlinear dynamic model of a CHP unit with AHP and bypass systems is proposed to reveal the internal principle of these two methods and their effect on CHP units, in which, the static unknown parameters are determined based on design data, and the dynamic ones are obtained by perturbation test. Simulation results show that the model can reveal the couplings of AHP and bypass system to the CHP unit, and provide model support for controller design. Secondly, a deep peak regulation control strategy is proposed to improve the peaking capacity of the CHP unit, in which, the GPC algorithm with feedforward-feedback structure is adopted to fundamentally solve the control problems of large delay and inertia on the boiler side, and overcome known disturbances on the turbine side; the ratio of the first stage pressure to the exhaust pressure from high pressure cylinder is selected as the controlled variable of high pressure bypass. The deep peak regulation process is divided into two stages: 1) CHP and AHP are used for heating when their heating capacity can meet the heat load requirements of residents, 2) as the unit load is further reduced, bypass mode is activated for heating when the steam flow entering the LPC is lower than its cooling flow. Simulation results show that the control strategy can meet the heat load requirements of residents when the unit is in deep peak regulation condition, the steam flow entering LPC can always be kept above its limit value (180t/h), the ratio of the first stage pressure to the exhaust pressure from high pressure cylinder can be always kept within a certain range, therefore, the safe and stable operation of the unit can be guaranteed. ZEKUN XIE is currently pursuing the Ph.D. degree in control theory and control engineering with North China Electric Power University, Beijing, China. He is mainly interested in condition detection and diagnosis of thermal system equipment, where he has made great achievements in this field. VOLUME 8, 2020