Academic Journal of Science and Technology 
ISSN: 2771-3032 | Vol. 15, No. 2, 2025 

 

132 

SSA‐based	TVOC	Concentration	Control	Method	for	
Storage	Equipment	
Chenchen Liu*, Jun Li, Ye Yuan 

College of Mechanical Engineering, Sichuan University of Science & Engineering, Sichuan, China 
* Corresponding author 

 

Abstract: The collection and storage of hazardous waste (HW) represent a critical phase in the full-process management of 
HW, where the control of total volatile organic compound (TVOC) concentrations in storage equipment directly impacts 
environmental safety and personnel health. Traditional manual control of UV lamps relies on empirical judgment, often suffering 
from delayed response, high energy consumption, and poor stability. To enhance the safety management of HW storage 
equipment and achieve precise, efficient TVOC concentration control, this study proposes an intelligent control method based 
on a PID algorithm optimized by the Sparrow Search Algorithm (SSA). This approach dynamically adjusts UV lamp intensity 
in response to fluctuations in liquid waste volatilization, maintaining TVOC levels stably below the emission standard threshold 
of 500 ppb. Stability tests demonstrate that the SSA-optimized PID parameters exhibit rapid convergence, achieving parameter 
optimization for step and sinusoidal responses within 13 and 18 iterations, respectively. Experimental results on TVOC 
concentration control show that the proposed method stabilizes TVOC levels within the range of 216–228 ppb within 27 minutes, 
with minimal sensitivity to environmental variations. 

Keywords: PID; TVOC concentration control; SSA. 
 

1. Introduction 
In hazardous waste storage equipment, liquid wastes such 

as waste engine oil and waste organic solvents generate 
TVOC (Total Volatile Organic Compounds). Excessive 
TVOC concentrations may lead to safety accidents and cause 
health issues, including allergic reactions, respiratory 
infections, headaches, eye irritation, and nausea.To assess 
indoor air quality, appropriate detection methods must be 
employed to measure TVOC concentrations. Traditional 
monitoring techniques include gas chromatography-mass 
spectrometry (GC-MS) and gas chromatography-
fluorescence detection (GC-FD). With technological 
advancements, real-time TVOC gas sensors have been 
developed to enable prompt control measures[1].Currently, 
the primary methods for reducing TVOC levels are:Activated 
carbon adsorption: Due to its porous structure, activated 
carbon exhibits strong adsorption capacity, effectively 
trapping organic particles.UV photocatalytic degradation: 
Photocatalysis is an efficient method for decomposing 
TVOCs. Photocatalysts release negative ions, promoting 
chemical reactions that break down TVOCs, with ultraviolet 
(UV) radiation being the most effective excitation source[2]. 

This study proposes a combined approach where UV light 
activates a photocatalyst to decompose TVOCs, while 
activated carbon adsorption further reduces concentrations to 
meet national emission standards (below 500 ppb). However, 
current UV lamp control relies on manual intervention, often 
leading to delayed activation or insufficient intensity, 

resulting in TVOC exceedances.To optimize UV lamp control, 
this paper introduces a PID (Proportional-Integral-Derivative) 
algorithm that automatically adjusts UV lamp intensity via 
variable resistance. 

However, traditional PID tuning methods depend on 
empirical manual adjustments, which are inefficient and may 
not adapt to varying environmental conditions. Moreover, as 
the system operates over time, UV lamp performance 
degradation reduces decomposition efficiency, causing 
unstable TVOC control and necessitating frequent manual 
recalibration.To address these limitations, this study employs 
the SSA (Sparrow Search Algorithm) for automatic PID 
parameter tuning, enabling adaptive and automated UV lamp 
control[3]. The SSA algorithm was selected based on its 
optimization capability, convergence speed, global search 
performance, and computational efficiency, ensuring robust 
and stable TVOC concentration management under dynamic 
conditions.  

2. Theoretical Basis 

2.1. PID Control Theory 
The PID controller, as a classic closed-loop control strategy, 

holds significant application value in the field of industrial 
control. It operates by comparing the error between the 
system's actual signal and the predefined target signal, then 
adjusts the control output through proportional (P), integral 
(I), and derivative (D) actions to achieve precise control of the 
target process[4]. The PID control flowchart is illustrated in 
Fig .1. 

 



 

133 

 

Figure 1. PID control basic flow chart 

 

In PID control, the proportional term (P) establishes a 
linear relationship between the output and the instantaneous 
error, thereby improving the system's response speed to 
deviations. The integral term (I) adjusts the output based on 
the accumulated error, effectively eliminating the steady-state 
error of the system. The derivative term (D) predicts the trend 
of the error by considering its rate of change, helping to 
reduce overshoot in the system.The mathematical expression 
of the PID is shown in Eq. 1: 

 

 
0

( )
( ) ( ) ( )

t

p i d

de t
u t K e t K e t dt K

dt
       (1) 

 
where e(t) represents the error between the setpoint and the 

actual value, while pK  , iK  , and dK   denote the 

proportional, integral, and derivative coefficients, 
respectively. 

2.2. SSA theory 
The Sparrow Search Algorithm (SSA) is an intelligent 

optimization algorithm inspired by the foraging behavior of 
sparrow populations, proposed by Xue and Shen in 2020[5]. 
SSA includes three roles: finder, follower, and watcher to 
achieve different behavior patterns. In SSA, discoverers 
typically constitute 20% of the population, while followers 
account for 80%. The transition between these roles is 
dynamic, depending on whether a sparrow demonstrates 
enhanced food-finding ability. The algorithmic model of SSA 
is illustrated in Fig. 2. 

 

 

Figure 2. Sparrow search algorithm model 

 

In the SSA algorithm, the information exchange, 
cooperation, and individual exploration among sparrows are 
simulated, continuously updating their positions to ultimately 
find the globally optimal fitness value, thereby solving the 

optimization problem. Each sparrow's position in the 
population represents a solution, and the more food at a 
position, the better the solution, i.e., the higher the fitness. 
Since the discoverers search for food for the sparrow 

input +

‐

Proportional adjustment

Differential adjustments

Integral adjustment

Measuring cell

Actuator
output

The search 
is wide

Good 
adaptability

Finder
（20%）

Followers
（80%）

Poor 
fitness

Vigilant Solution 
space

Search 
extensively

Search near 
the finder

Dynamic 
swapping of 

identities



 

134 

population and provide foraging directions for the followers, 
the discoverers have a larger foraging range. During the 
iterative process of the algorithm, the position update function 
for the discoverers is given by Eq. 2. 

 

 
, 21

max,

, 2

,  

,  

t
i jt

i j

t
i j

i
X exp R ST

iterX

X Q L R ST



  
      

   

 (2) 

 
Where t represents the current iteration count; itermax is the 

maximum iteration count; α is a random number within the 
range (0,1]; Q is a random number following a normal 
distribution; L is a 1×d matrix where every element is 1; 
R2(R2∈[0,1]) is the warning value; and ST(ST∈[0.5,1]) is the 
safety threshold. 

When R2＜ST, it indicates that the nearby area is safe, and 
the discoverers will conduct an extensive search.When R2 ≥ 
ST, it means there are predators in the vicinity, making the 
area unsafe, and all sparrows must immediately fly to a safe 
position. 

The followers have weaker foraging abilities but constantly 
observe the discoverers. When a discoverer finds more food, 
the followers will leave their current positions and follow the 
discoverer to forage. The function for the follower position 
update is shown in Eq. 3. 

 

,

21
,

1 1
,

  / 2

     

t t
worst i j

t
i j

t t t
P i j P

X X
Q exp i n

iX

X X X A L else



  

  
       

    

     ，

，

(3) 

 
Where Xp represents the current global best position of the 

discoverer; Xworst denotes the current global worst position; A 
is a 1×d matrix where each element is either 1 or -1; When i  
(n/2), it indicates that the followers with poorer fitness values 
in the population will fail to obtain food and must fly to 
positions with more food. 

It is assumed that scouts account for 20% of the population 
in the model. When the scouts detect a predator, the sparrow 
population's alarm mechanism will be triggered, as shown in 
Eq. 4. 

 

 

,

1
, ,

,

        ,

   ,

t t t
best i j best i g

t t t
i j i j worstt

i j i g

i w

X X X f f

X X X
X K f f

f f







    
  

        

 (4) 

 
Where Xbest represents the position with the most food in 

the population; β is a random movement control parameter 
with a variance of 1 and a mean of 0; K(K∈[-1,1]) is the step-
size control coefficient, indicating the direction of the 
sparrow's movement; fi denotes the fitness value of the i-th 
sparrow individual; fw represents the current global minimum 
fitness value; fg is the current global maximum fitness value; 
and ε is a minimal constant to prevent division by zero in the 
formula. 

When fi  fg, it means the current sparrow is at the edge of 
the population, making it more vulnerable to predator attacks, 
so it needs to fly toward the center of the population. When fi 
= fg, it indicates that the scout sparrow has detected danger 
and must move closer to other individuals to reduce the risk 
of predation. 

The basic workflow of the SSA algorithm is illustrated in 
Fig. 3-6. 

 

 

Figure 3. Basic flow chart of SSA algorithm 

 

The Sparrow Search Algorithm (SSA) exhibits strong 
optimization capabilities and fast convergence speed. By 
employing the SSA, the optimal parameters Kp, Ki, and Kd in 
PID control can be quickly determined, thereby enabling 
precise regulation of ultraviolet lamp intensity and activation 
time to maintain TVOC concentration below the standard 
limit. 

3. Determination of the Function 

3.1. Control the model transfer function 
Transfer function modeling is a critical step in control 

system design, as it accurately characterizes the dynamic 
relationship between system inputs and outputs. In the UV-
based TVOC concentration control system, the TVOC sensor 
monitors the ambient concentration in real time and compares 
it with the target value to generate an error signal. The PID 
controller then calculates and outputs corresponding UV 
intensity adjustment commands based on this error, enabling 
dynamic regulation of TVOC concentration[6]. The UV lamp 
control model is illustrated in Fig. 4. 

 

开始

初始化种群和参数

预警机制更新部分个
体位置

跟新发现者和追随者
个体位置

计算种群适应度函数

更新适应度值，获取
更优位置

是否满足优化终止条件

输出最优[K，α]

结束

N

Y

Begin

Initialize populations 
and parameters

Calculate the fitness 
function

Update the location of 
finders and tracers

The early warning 
mechanism updates the 

individual location

Update fitness values 
for better locations

Whether the 
optimization termination 

conditions are met

The output is 
optimal [K,α]

End



 

135 

 

Figure 4. UV lamp control model 

 

The intensity and activation duration of UV light directly 
influence the degradation rate of TVOCs. Given the 
oscillatory and delayed dynamic characteristics observed in 
the UV control system, The second-order differential 
equations shown in Eq. 5 are established to establish the 
transfer function to describe the dynamic behavior of the 
system. 

 

 
2

2

2

( ) ( )
2 ( ) ( )

d C t dC t
T T C t K I t

dt dt
     (5) 

 
Here, K represents the system gain, reflecting the static 

influence of UV intensity on TVOC concentration; T is the 
time constant, determining the system's response speed; ξ is 
the damping ratio, governing the system's oscillatory 
properties; C denotes TVOC concentration; and I(t) 
represents UV lamp intensity. 

The Laplace transform is performed on the above 
differential equations to obtain the frequency-domain 
equation as shown in Eq. 6. 

 

 2 2 ( ) 2 ( ) ( ) ( )T s C s TsC s C s K I s       (6) 

 
The transfer function as shown in Eq. 7 is obtained. 
 

 2 2

( )
( )

( ) 2 1

C s K
G s

I s T s Ts
 

 
    (7) 

 
To determine the parameters of the transfer function (K, ξ, 

T), it is essential to understand the mechanisms of TVOC gas 
volatilization and UV lamp-induced degradation. In this study, 
experiments on TVOC volatilization and degradation were 
conducted in a closed environment of a storage device. The 
volatilization conditions included a temperature range of 23–
25°C and a volatilization area of approximately 0.24 m².The 
TVOC concentration released from waste liquids (waste 
organic solutions and waste lubricating oil) was measured 
over 300 minutes after the liquids were introduced into the 
waste tank. Data was collected at a frequency of 1 minute per 
sample. The fitted trend of TVOC gas concentration released 
from the waste liquids is shown in Fig. 5. 

 

 

Figure 5. Trend of volatile TVOC gas concentration in waste liquid 

 

As shown in Fig. 5, the TVOC concentration increased from 47 ppb to 882 ppb over 300 minutes, with the growth 

Expectations
R（s ）

+
-

error
E（s ）

C（s）

Controller 
output
F（s ）

Environmental 
interference

D（s ）

-

+
2 2 2 1

K

s s  

TVOC
Concentration

C（s ）

Ultraviolet 
lamp system

PID
controller

0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300
0

100

200

300

400

500

600

700

800

900

T
V
O
C
 
c
o
n
c
e
n
t
r
a
t
i
o
n
（

p
p
b
）

Time（min）



 

136 

rate gradually slowing down. This phenomenon occurs 
because organic gases inherently have a saturated vapor 
pressure—the higher the gas concentration, the lower its 
saturated vapor pressure, making further volatilization more 
difficult. 

Three UV lamp intensity levels—20 W/m³, 30 W/m³, and 
50 W/m³—were used to degrade TVOC with an initial 
concentration of 600 ppb (while waste liquid volatilization 
continued). The fitted TVOC concentration curves under 
these conditions are presented in Fig. 6. 

 

 

Figure 6. Effect of three intensities of UV lamp on TVOC degradation 

 

As illustrated in Fig. 6, UV lamps with intensities of 20 
W/m³, 30 W/m³, and 50 W/m³ reduced the TVOC 
concentration from 600 ppb to 226 ppb, 221 ppb, and 186 ppb, 
respectively, over 300 minutes. Lower UV intensities resulted 
in slower concentration reduction rates, with all rates 
eventually approaching zero. This occurs because as the 
concentration decreases, fewer molecules remain available 
for decomposition, slowing the degradation rate. 
Simultaneously, the reduced saturated vapor pressure of the 
waste liquid gradually increases its volatilization rate, 
ultimately leading to concentration stabilization. 

Using the experimental data, the functional relationship 
between input and output signals was fitted through system 
identification methods, with continuous refinement of the 
transfer function parameters. The derived transfer function for 
the UV control system is as follows: 

 

 
2

33
( )

17 28 1
G s

s s


 
             (8) 

 

3.2. Optimal solution fitness function 
The selection of the fitness function is a critical step in 

intelligent algorithms, as it determines the direction of 
optimal solution search and evaluates the quality of solutions 
for a given problem. Based on fitness values, the algorithm 
prioritizes solutions and allocates exploration opportunities. 
When designing a fitness function, optimization objectives 
and system constraints must be clearly defined. 

For the UV lamp-based TVOC concentration control 
system, the objective function should holistically consider 
control accuracy, energy consumption, and system stability to 
maintain TVOC levels as close as possible to the setpoint with 

minimal energy expenditure. In this study, control accuracy is 
quantified by the deviation between the measured TVOC 
concentration and the target value, as expressed in Eq. 9. 

 

 2
1 0

( ( ) )
T

setJ C t C dt          (9) 

 
Where C(t) represents the real-time TVOC concentration 

measured by the sensor, Cset is the target concentration value, 
and T denotes the control duration. 

Energy consumption is quantified by the total power usage 
of the UV lamp, as expressed in Eq. 10. 

 

 2 0
( )

T
J P t dt              (10) 

 
P(t) represents the UV lamp's power consumption, which 

is typically proportional to its intensity I(t). 
System stability is evaluated based on the fluctuation of 

control signals, primarily to prevent excessive variations in 
UV lamp intensity that could compromise its lifespan, as 
shown in Eq. 11. 

 

 2
3 0

( )
( )

T dI t
J dt

dt
            (11) 

 
A weighted sum of control accuracy, energy consumption, 

and system stability is discretized to derive Eq. 12. 
 

2

2
1 2 3

1 1 1

( ) ( 1)
( ( ) ) ( )

N N N

set
k k k

I k I k
J w C k C w P k w

t  

        
   (12) 

0 20 40 60 80 100 120 140 160 180 200 220 240 260 280 300
150

200

250

300

350

400

450

500

550

600

650

T
V
O
C
 
c
o
n
c
e
n
t
r
a
t
i
o
n
（

p
p
b
）

Time（min）

 I1=20 W/m
3

 I2=30 W/m
3

 I3=50 W/m
3



 

137 

where w₁, w₂, and w₃ are the weighting coefficients for 
control accuracy, energy consumption, and system stability, 
respectively; N is the number of time steps; and Δt is the time 
step size. 

4. Experimental Results 
The SSA algorithm is employed to automatically optimize 

the PID control parameters for UV lamp intensity. In the SSA 
configuration, the population size is set to 50 sparrows, with 
a maximum iteration count of 100. The discoverers account 

for 20% of the total population, followers make up 80%, and 
scouts constitute 20% to prevent the system from converging 
to local optima. 

In this study, a unit step input signal is used to simulate 
scenarios where TVOC concentration rapidly increases from 
zero, thereby testing the PID controller's performance. The 
objective is to maintain TVOC concentration around 500 ppb 
while minimizing UV lamp energy consumption. The 
simulation results for the unit step response are presented in 
Fig. 7, 8. 

 

 

Figure 7. Convergence plot of unit step input signal response 

 

 

Figure 8. TVOC concentration control chart of unit step input signal 

 

As shown in Fig. 7, the fitness value progressively 
approaches the optimal solution with increasing iterations. 
When the iteration count reaches 13, the algorithm achieves 
the optimal fitness value of 534,341, and the fitness curve 
begins to converge, demonstrating the SSA algorithm's rapid 
convergence speed. 

The optimized PID control parameters obtained are 
Kp=0.36179,Ki=0.022452,Kd=0.39605. After applying these 
parameters to the PID-based concentration control system, the 
TVOC concentration control simulation results are illustrated 
in Fig. 8. When the concentration rapidly exceeds 500 ppb 
and continues to rise at a certain rate, it stabilizes within 40 
minutes. The minor fluctuations in the actual concentration 
are attributed to the 1-minute sampling frequency and the 

inherent response delay of UV lamps in reducing the 
concentration. These results confirm the effectiveness of the 
SSA-optimized PID parameters in controlling TVOC 
concentration. 

5. Summary 
This study addresses the TVOC concentration control 

problem in hazardous waste storage equipment by designing 
an SSA-optimized PID control system. Through theoretical 
analysis, algorithmic optimization, and experimental 
validation, the following conclusions are drawn.The SSA-
optimized PID parameters demonstrate rapid convergence, 
completing step response parameter optimization within 13 
iterations and sinusoidal response optimization within 18 



 

138 

iterations - showing significant efficiency improvement over 
traditional manual tuning. The optimized PID parameters 
enable TVOC concentration to stabilize at the target value of 
500 ppb within 40 minutes, proving the UV lamp control 
system's strong adaptability to different control states. 

References 
[1] Palmisani J, Di Gilio A, Viana M, et al. Indoor air quality 

evaluation in oncology units at two European hospitals: Low-
cost sensors for TVOCs, PM2.5 and CO2 real-time 
monitoring[J/OL]. Building and Environment, 2021, 205: 
108237. https://doi.org/10.1016/j.buildenv.2021.108237 

[2] Kang Y, Jo H H, Kim S. Effects of UV degradation on building 
materials with emphasis on microplastic generation 
potential[J/OL]. Journal of Hazardous Materials, 2025, 483: 
136521. https://doi.org/10.1016/j.jhazmat.2024.136521 

[3] Zhang Y, Wang X, Qian X, et al. Research on load frequency 
control of new power system based on improved sparrow 
search algorithm[J/OL]. Measurement: Sensors, 2024: 101432. 
https://doi.org/10.1016/j.measen.2024.101432 

[4] Cantera-Cantera L A, Maya-Rodríguez M C, Palomino-
Resendiz S I, et al. Level and flow systems identification of an 
industrial processes module by LSOD method for PID 
controllers design[J/OL]. Results in Engineering, 2025, 25: 
104347. https://doi.org/10.1016/j.rineng.2025.104347 

[5] Xue J, Shen B. A novel swarm intelligence optimization 
approach: sparrow search algorithm[J/OL]. Systems Science & 
Control Engineering, 2020, 8(1): 22-34. 
https://doi.org/10.1080/21642583.2019.1708830 

[6] Parsa Z, Dhib R, Mehrvar M. Multi-Loop PID Controller 
Design for PVA Degradation in a Tubular UV/H2O2 
Photoreactor[J/OL]. IFAC-PapersOnLine, 2024, 58(14): 736-
741. https://doi.org/10.1016/j.ifacol.2024.08.425