Introduction

The world-wide search for sustainable energy solutions has led to the increased look for potential bio fuels for CI engines with biodiesel usually being a possible renewable fuels option28,29,30,31. According to the latest figures, global biodiesel output was estimated at 34 billion liters in 2023, illustrating new trends for green fuel replacement1,2,3. Comparing to the other types of biodiesel feed stocks Annona biodiesel appears to be most attractive and favourable feed stock for biodiesel production due to its local availability and advantageous combustion properties. The application of ML into the optimization of the engines has changed the way of improving the engine performance4. Research has shown that the ML directed optimization has possibility of providing an improvement of more than 2% in engine performance parameters over conventional CFD approaches. This advancement is particularly significant when premised on the issues that CI engines struggled to overcome while using biodiesel fuels particularly, due to its high viscosity and low calorific value relative to standard diesel6,7.

Biodiesel with Annona in CI engines has shown positive results when experimented with different blend ratios of Annona biodiesel. Closely related, experimental results indicate that the A20 (20% Annona biodiesel (Annona spp.) and 80% diesel) surrogate fuel blend displayed relatively sound performance parameter with the highest brake thermal efficiency and miserly emissions8. Laboratory analysis of biodiesel blends from Annona revealed that the biodiesel blends had several performance parameters and that blend ratios affected performance while others suggested that brake thermal efficiency ranged from 17 to 24% in some operational conditions9. Therefore, machine learning algorithms especially the XG Boost and Random Forest models have remarkable accuracy in the prediction of the said parameters of the engine performance10. These techniques of ML need a lot of appreciation for helping in fine tuning some of the most sensitive parameters which are intimately related to the engine parameters like injection timing, compression ratio and the fuel injection pressure which all together contribute towards enhancing the ‘power factor’ and, therefore, the overall efficiency of the engine. Experiments and real-world simulations of the latter implementations of optimizing ML approaches claim upper fuel savings of up to 8.256% with a guarantee of intact substantive aspects of air performance11,12. Table 1 shows that Summary of various studies.

Table 1 Summary of various studies.
Full size table

Especially, the difficultly in maintaining the pother power factor for the CI engines operating on biodiesel has been a subject of interest to the researchers. There is a research evidence showing that biodiesel blends generally raise the brake specific fuel consumption (BSFC) as the biodiesel content increases but this phenomenon can be well controlled as found in present investigation through ML optimized engine parameters13. Efforts are notable for real time optimization of the engine parameters utilizing biodiesel fuels through the incorporation of complex ML processing hence has enhanced the power factor despite considered as a major technical concern14. The studies have shown that Annona biodiesel blends can perform as good as the traditional diesel if processed with optimization using the ML methods. It has been reported that optimized biodiesel blends have a brake thermal efficiency of 2–3% of conventional diesel while at the same time possessing a relatively low emissions’ output15. This optimization is realized by having very tiny control of different characteristics of the engine which by the use of the ML algorithm one is able to process and analyze very complicated operation data of the engine at real time16.

The use of ML methodologies in engine optimization has shown effectiveness in time and operations savings for performance upgrade. SEO tactics in the past crucially involve numerous parametric tests to step-up a product as well as to create an efficient web-page, however the ML methods are able to evaluate basic parameters of the product accurately and without lots of experiments. This efficiency in optimization has ensured that it is possible to obtain enhanced power factors with biodiesel fuel while at the same time retaining the environmental advantages of biodiesel fuel17. Recent development in the area made it possible to use advanced ML algorithms for engine parameter optimization. The application of artificial intelligent techniques, especially artificial neural networks and support vector regression requisite; better predictions of various performance parameters of the engine have been estimated. More complex ML applications have proven especially useful in enhancing power factor with regard to the properties of biodiesel blends derived from Annona seeds18.

The effect overall of the Annona biodiesel on the environment as well as fuel consumption through the engine parameter adjustment using the ML technique is encouraging. It was established that CO and unburned HC emissions were lower than those obtained with conventional diesel while having comparable power output19. The application of the ML methods has enabled the engine parameters to be optimised and counter some of the major drawbacks of biodiesel technology such as, rising levels of NOx and enhanced fuel consumption. The practical feasibility of the new approach using ML-optimized engine systems is a concept that has now become clear. MO investigations suggest that the initial cost of implementing ML is easily recovered from the faster fuel consumption and minimal servicing expenses20. The fact that the designs can sustain the power factor at its optimum maximum even when employed to use re-newable fuel source is a positive step towards the solution for efficient transportation21.

The following are the main research gaps that cut across the current state of ML-driven optimization of CI engines using Annona biodiesel: The rarity of models identifying AI’s application in the integration of nano-fuels and different concentrations of nanoparticles is one major deficit this work has identified. Diesel engine has attracted little research specifically geared toward utilization of meta-heuristic techniques for optimization, and biodiesel is quite an uncharted area in this context. Further research is called for on the use of multiple ML techniques to perform optimization tasks of both performance and emissions at the same time1. This partially applied research gap holds current studies on the different ML algorithms and their capability of comparing blended biodiesel–alcohol and their impacts on engines’ performance when there is no detailed computational solution involving different ratios of biodiesel-alcohol blend. Furthermore, the integration of fuzzy logic with particle swarm optimization for improving engine performance is not studied enough.

The objective of this study is to examine common power quality problems, identify their root causes, and explore the techniques, devices, and tactics employed to mitigate their effects on systems. The next thing that will be done is to provide an analysis of the data that were obtained from the two separate sources of information on diesel engine power quality studies. In a generator powered by a diesel engine, noise and vibration are extremely important components. In a generator, the engine is responsible for producing the highest possible levels of noise and vibration intensity depending on the frequency. For the purpose of this experiment, a conventional diesel-biodiesel combination was utilised to provide power to an unmodified diesel engine at one of the following rotational speeds: 1200 revolutions per minute (rpm), 1500 rpm, 1800 rpm, 2100 rpm, and 2400 rpm. Anonna biodiesel is used in lieu of 20% of the conventional diesel fuel throughout each and every one of the tests. The ongoing inquiry aims to determine several engine parameters, such as power, torque, noise, and vibration. This will be done by analysing the impact of input elements, including fuel mix, engine speed, and load.

Materials and methods

Biodiesel preparation

It is for the purpose of harvesting the fruit that many tropical nations cultivate the Annona squamosa Lin. plant, which is also commonly referred to as the custard apple. It is a little tree that is indigenous to tropical America and the West Indies. The A. squamosa is a species of the genus A. It has little leaves and is almost always in a state of evergreenness. It is for the purpose of gathering the fruit of the custard apple tree that it is planted over the entirety of India. The cultivation of a custard apple would typically include the usage of an entire species of Annona on the plant. The annona plant is known by a variety of various regional names in India. Some of these names are sita pandu, sarifa, sitaphal, sharifa, and custard apple. Additional names include sharifa. Depending on the species, an A. squamosa tree may produce anywhere from 34 to 45 kilogrammes of fruit per year. A rose-golden colour can be seen in the pulp of the fruit, which is also fairly delicious and exceptionally tender. In addition to having a smoky brown colour, the skin of the fruit is apparent with depressions that give it the appearance of being quilted. During one point in time, it was also referred to by the name sweetsop. As an additional point of interest, the kernels of seeds and fruits are utilised in the manufacturing of both soap and oil. Figure 1 shows that Custard apple.

Fig. 1
Fig. 1The alternative text for this image may have been generated using AI.
Full size image

Custard apple.

Fig. 2
Fig. 2The alternative text for this image may have been generated using AI.
Full size image

Custard apple pulp with cross section showing seeds.

Figure 2 shows that Custard apple pulp with cross section showing seeds Glycerol and esters are the products of the chemical reaction known as transesterification, which involves passing an oil or fat through an alcohol in the presence of a catalyst. Following the reaction of the alcohol with the fatty acids, the mono-alkyl ester, also known as biodiesel, and crude glycerol are produced. Methanol is the alcohol that is utilized in the majority of the biodiesel synthesis process, and either potassium or sodium hydroxide is employed to catalyze the reaction into a base. The reaction of raw oil and methanol results in the production of oil methyl ester, which is the end product of the process of transesterification.

During the process of esterification, vegetable oil is subjected to a chemical reaction at a temperature that is held constant while the oil is continuously stirred. Because the esterification reaction cannot take place in the esterification process without a catalyst, catalysts are extremely critical for both acid and alkaline esterification. It’s possible that the amount of catalyst that’s needed for this reaction will change based on the type of fatty acid that’s found in the vegetable oil. The alkali-catalyzed approach completes the transesterification reaction far more quickly than the acid-catalyzed process does, making it the most economically viable option. The esterification has brought about a significant change in the vegetable oil by bringing about a reduction in its viscosity. The biodiesel crop, which was formed after acid and alkaline esterification, delivers the well outcomes in lowering its viscosity, and by confirming with the features of diesel, it generates results that are extremely comparable to the results produced by diesel. Various properties of Annona biodiesl and diesel is shown in the Table 2.

Table 2 Properties of annona biοdiеsеl аnd diеsеl.
Full size table

Experimental methods

In this particular investigation, experiments were conducted three times at each of the five separate engine speeds that were as follows: 1200 revolutions per minute (rpm), 1500 rpm, 1800 rpm, 2100 rpm, and 2400 rpm. The annona fuel mix has been manufactured at this time in order to get ready for these experiments. In order to guarantee that all possible modes of engine operation are covered, experiments are carried out with the engine working at a variety of speeds and under full load. data are obtained of the torque, power, fuel consumption, and SFC throughout each test. Additionally, data are collected of the temperature, pressure, and humidity of the environment that is around the vehicle. Following this, the readings are entered into a computer programme.

Test engine

The use of a test engine with four cylinders and water cooling allows for the collection of experimental data, specifications of the test engine was given in the below Table 3. The engine was put through its paces on a DC (Direct Current) dynamometer (brand name: Kemsan), which has the capacity to generate 5.2 kW of electricity. In the course of this investigation, a Kistler Rotor type 4550 A Torque measuring System was utilized so that the engine torque values could be measured. Accuracy within 0.05% is provided by this sensor. Additionally, with the assistance of a magnetic system, it is feasible to determine the torque values derived from the shaft. Specifications of the measuring equipment’s was shown in the Table 4.

Table 3 Engine specifications.
Full size table
Table 4 Specifications of the measuring equipment’s.
Full size table

Power quality monitoring

The power quality of the electric network is put at risk by harmonics, and the operation of equipment is disrupted. In order to increase the quality of the power, it is important to compensate for harmonic signals. Harmonic analysis may be done in a number of different ways to enhance power quality. Some examples of these approaches are the Fast Fourier Transform (FFT), singular value decomposition (SVD), and artificial neural network (ANN).

The STATCOM approach is used throughout this current study as the fundamental mitigation strategy for controlling power quality issues.When it comes to diesel engine generators, STATCOM may be an option worth considering for addressing power quality concerns. STATCOM is a piece of power electronics equipment that works on the fundamental idea of injecting or absorbing reactive current. In order to combine a diesel engine generator with a power electronic interface, additional converters and power conditioning devices are necessary. The STATCOM unit was primarily developed for the purpose of providing reactive power compensation to the load. Additionally, it has an inverter that is equipped with a DC link capacitor. It is responsible for causing the STATCOM to supply actual power and reactive power adjustment after receiving control pulses that have been created using a modified version of the Icos algorithm. The STATCOM that is utilized for the biodiesel-powered diesel generator is seen in Fig. 3.

Fig. 3
Fig. 3The alternative text for this image may have been generated using AI.
Full size image

Schematic of the three-phase grid system with the STATCOM interface for renewable energy source.

STATCOM is a tool that may be very helpful in mitigating the power quality difficulties that occur in diesel engine generators. It is a device that is connected to the FACTS (Flexible AC Transmission System) transmission system in a shunt way and is one of such devices. A coupling transformer is used to provide a connection between the VSC and the system bus in this design. The amplitude of the VSC output has an effect on the reactive power. Under conditions of overload, STATCOM will inject reactive power into the system. However, if the terminal voltage (VDC) is lower than the system voltage, which indicates that it is experiencing under voltage, STATCOM will absorb reactive power. The STATCOM arrangement of the diesel generator that is powered by biodiesel is seen in Fig. 4.

Fig. 4
Fig. 4The alternative text for this image may have been generated using AI.
Full size image

DG set configuration which consists of a diesel engine driven squirrel cage induction generator along with a Distributed Static Compensator(DSTATCOM)23.

Vibration and nosie monitoring

Accelerometers were used during the trials to collect vibration data from the engine block. The measurements were taken at a sample frequency of 51.2 kHz, and the results from each experiment were collected for a duration of ten seconds.

Three accelerometers are required in order to generate the necessary engine vibration signals. Accelerometers were positioned in three different directions: vertically (x), laterally (y), and longitudinally (z), and vibration signals were sent to a switchboard from each of these accelerometers. An analog-to-digital (A/D) converter received the switchboard output signals before being processed further. Following that, the data recording was accomplished by connecting the output data connection to the USB port of a computer. Table 5 shows that Specifications of noise and vibration measuring instruments.

Table 5 Specifications of noise and vibration measuring instruments22.
Full size table

Root mean square (RMS) of acceleration signal

RMS is defined by Eq. (1) with the number of points within a time period, N, and acceleration data x(tk) in time of tk;

$${\text{a}}_{{\text{W}}} = \sqrt {\frac{1}{T}} \mathop \smallint \limits_{0}^{T} a_{W}^{2} \left( t \right)~dt$$
(1)

This amount serves as a criterion for comparing the vibration that occurs under a variety of different settings. The root mean square, or RMS, is proportional to the amount of energy that is included in vibration signals, which is dependent on the amplitude of the signal. As a result of the RMS being computed according to the square of the vibration data, abrupt peaks are automatically given more weight in the calculation.

Result and discussion

Power quality of the engine

The actions through which the waveform profile of a power system voltage and current are compelled to remain in the sinusoidal waveform are known as power quality, and power quality encompasses the actions. Incorporation of SR in a normal power system provides the system capability to counter variability, over- strategy and error in load and generation modeling. Overall, this suggests that the system is capable of addressing such problems a view deduced from the overall results below. The term ‘spinning reserve’ abbreviated as ‘SR’ is defined as the capability of a generator to provide a supply greater than rated, with a fundamental frequency, when additional torque is put on exercise.Despite the correction of power factor and current harmonics, through a specific power device namely the STATCOM, a shunt connected device, there is a general enhancement in power quality. This improvement is attributable to the use of the STATCOM as this study has shown. Moreover it may also filter the output, control voltage that appears on the distribution bus and act as load leveling device. Indeed all of the above additional functions are achievable. Through adding a part of the harmonic waveform in phase opposition and generated from the load STATCOM balances the detrimental effects that are as a result of harmonics.

Fig. 5
Fig. 5The alternative text for this image may have been generated using AI.
Full size image

Changes of power with fuel and speed.

Fig. 6
Fig. 6The alternative text for this image may have been generated using AI.
Full size image

Changes of power with fuel and speed.

To compare the both Figs. 5 and 6, torque and power that the test engine produced when fuelled with diesel and that time when fuelled with B20 biodiesel, the graph which is shown in the figures, five and six. Concerning the determination of power and torque that each fuel produces, an eddy current dynamometer is used in the evaluation process. Additional testing on the engine which has installed SATACOM electronic control is carried out to assess its power and torque capacity. The biodiesel Annona B20 attained its best torque, between 2100 and 2400 revolutions per minute (rpm) and maximum power of 1,2400 kW at 2400 rpm. The diesel fuel can deliver the greatest power of 4.8 kW at 2400 revolutions per minutes and torque of 10 Nm in the range of 2100–2400 revolutions per minutes. Some observations made recently indicate that the torque and power that can by gotten from biodiesel is slightly lesser than that gotten from diesel fuel. This time the dip in the torque is shown between 2100 and 2400 RPM though the trend can be observed again between the two RPMs. Diesel fuel can burn with higher speed than gasoline due to higher calorific value and relatively low viscosity of the fuel. This enables a larger mass of fuel to be injected into the cylinder at high engine speed without occurrence of what is called ‘diesel knock’. This in turn results to an increase in power and also the torque throughout the system. The increase in the value of CN is accompanied by a change in the time at which the cylinder has the maximum pressure as well as the decrease in the ignition delay. This in turn ensures that there is effective combustion of fuel and air mixture because of right photons increase. When the crankshaft requires a particular mode of rotation, it is possible to create some details that exhibit enhanced power. This is because of the fact above state. Despite of tis, WVO biodiesel showed no increase in power and torque as opposite to diesel. This was however attributed to the fact that the fuel was viscous and its calorific value was lower than the expected, they could not therefore cause the required effects. Also, it has been noted that the diesel generator that is worked in SATACOM current as well as power and torque in better way than other generators at each and every range of the engine speeds. This is so since the generator addresses problems such as power factor as well as current harmonics hence enhancing the quality of power output. It also raises the idea that it may also filter the output, it may control the voltage at the distribution bus and finally control the load.

Vibration of the engine

It is very clear from the Fig. 7, that the vibration depends upon the type of fuel and the speed of the engine. Consequently, it was established that the frequencies that were dominant were doing so at a rate that is double that of the healthy running engine. This was the case since the movement of the engine pistons both up and down gives the most vibrations. From the experimental four-stroke, four-cylinder diesel engine, the fundamental frequencies were observed at 40 Hz/ 1200 rpm, 50 Hz/ 1500 rpm, 60 Hz/ 1800 rpm, 70 Hz/ 2100 rpm and 80 Hz/ 2400 rpm. All of these frequencies were measured in the vertical ‘z’ direction. As the engine speed intensifies, vibration level raised along with each of the test fuels included in the study. This was the case of all test fuels used in the study. It is also observed that the greater the increase in engine speed the greater is the increase in oscillations but the total magnitude of vibration is getting smaller.

Fig. 7
Fig. 7The alternative text for this image may have been generated using AI.
Full size image

Change of vibration with fuel type and speed.

To calculate the variations in the levels and to identify which of the test fuels has a resultant effect, the RMS value is recorded together with the changes observed. Hence, the research indicates that the level of vibration that is generated by the engine is directly related to the rate of operation of the engine. This is owing to the fact that at high engine speeds there are large inertial forces within the engine. The oscillation of the engine block is eliminated when each test fuel is being used and the performance difference becomes significant when the engine is run at a faster rate. That this effect is in fact there, is a matter of primary importance to take into account. It was also shown that, although the 20% biodiesel blend exhibited increased vibration acceleration than diesel at 1200 and 1800 rpm, it reduced the vibration acceleration at other rpm. Even if it was not comparing diesel, this was the case. To compare, it is requisite to record vibratory characteristic at an engine speed of 2400 rpm to isolate the fundamental variation that exist in between diesel and biodiesel blend. To my mind, the given can be attributed to the peculiarities of the fuel and the manner, in which it determines the process of combustion occurring inside the cylinders. Amongst all of these characteristics one can highlight the biggest peak pressure increase rate. Others examples of properties belonging to this category are the time of combustion and the delay time. It may be attributed to the Global ignition delay may possibly be lowered in employing biodiesel since it generally exhibited a higher level of oxygen content than diesel.The combustion contributed to the improved quantity of oxygen existing in the annona. This in its turn of course made the engine run much smoother due to the improved combustion that took place. This lead to the reduction of the present vibration level, for it was its aftermath as highlighted by Equation 2.

Nosie of the engine

Fig. 8
Fig. 8The alternative text for this image may have been generated using AI.
Full size image

Change of noise with fuel type and speed.

In Fig. 8 below is an example of how the level of engine noise varies with the rotational speed of the engine. The variations which form the basis for engine vibration are the causes of engine noise. Therefore, the amount of vibration that is created by this engine corresponds with the amount of sound which is created by the engine. From the tests it has been established and concluded that there is direct relationship between the amount of sound produced and the rate of speed of the engine since the vibration of the engine block increases proportionally with the rate of speed of the engine. This is evidenced by the fact that the number of throws has also increased with the engine speed of the sound produced. Since both engine body vibration and combustion noise are influences of engine sound, less vibration has been observed to lead to less sound in biodiesel run engines. This is because both of them are source of engine sound can be explained. The step of combustion that occurs during operation of for example a diesel engine is probably the most dominant source of noise that arises from the engine.The time spent in identification mode can be hereby minimised thus reducing equally the noise and vibration associated with the engine operation. Because of higher oxygen content of B20 therefore it has been found that use of annona can greatly reduce the amount of noise produced by the engine regardless the speed. This can be due to the enhanced combustion process that happens to be critical in enhancement of fuel combustibility.

Machine learning models for power and torque prediction

In this study, we employed four machine learning (ML) algorithms—Multiple Linear Regression (MLR), Gaussian Process Regression (GPR), Support Vector Machine Regression (SVMR), and Random Forest Regression (RF)—to predict the power and torque of an Annona biodiesel-powered engine under three different fuel conditions: Diesel, B20, and B20 Statcom. The performance of these models was evaluated using the R2 score, Mean Absolute Error (MAE), and Mean Squared Error (MSE), with the results summarized in Table 3. MLR showed strong predictive capability for both power and torque, achieving R2 scores of 0.883 and 0.947, respectively, indicating that the model effectively captured the variance in the data. GPR, on the other hand, underperformed, particularly in power prediction, with a negative R2 score of -0.042, suggesting that it failed to model the relationship between input variables and power accurately. SVMR provided moderate accuracy, with R2 scores of 0.722 for power and 0.721 for torque, but the relatively high error metrics indicated room for improvement. RF emerged as the most effective model, with R2 scores of 0.888 for power and 0.966 for torque, demonstrating its robustness in accurately predicting these performance metrics.

The performance metrics for different ML algorithms, as presented in Table 6, provide a detailed comparison of their accuracy in predicting power and torque. This table highlights the strengths and weaknesses of each approach, showing that while MLR and RF performed well, GPR was less effective. The practical implications of these findings are significant. Accurate ML predictions can optimize diesel engine performance by enabling better tuning and nanoparticle utilization, ultimately leading to more efficient engines. Furthermore, improved predictive accuracy can help reduce emissions, contributing to environmental sustainability. The methodologies used in this study can also be applied to other engines and fuel types, broadening the scope of ML applications in thermal system optimization.

Table 6 Performance metrics for different ML algorithms.
Full size table

Figures 9 and 10, which plot the predicted versus actual power values, along with Figs. 3 and 4 for torque, visually demonstrate the accuracy of the models. These plots reveal how closely the predicted values align with the actual data, with RF showing the closest alignment, indicating its superior predictive capability. Overall, the integration of ML algorithms in this study not only enhances our understanding of biodiesel engine performance but also offers practical tools for optimizing engine operations and reducing environmental impact.

Fig. 9
Fig. 9The alternative text for this image may have been generated using AI.
Full size image

Relationship between predicted and actual power for different machine learning algorithm.

Fig. 10
Fig. 10The alternative text for this image may have been generated using AI.
Full size image

Relationship between predicted and actual torque for different machine learning algorithm.

Conclusions

In the current research, power quality, noise, and vibration distributed generator systems were investigated utilizing biodiesel in a CI engine. It has been noticed that the STATCOM approach is effective in reducing current harmonics and power factor problems. Additionally, it functions as a load balancer, a voltage regulator at the distribution bus, and a filter. STATCOM coupled with an altered version of the ICOS algorithm has the potential to alleviate power quality concerns in CI engines that run on biodiesel. Because it is capable of providing both active and reactive power compensation and has superior functional qualities, quicker performance, a smaller size, cost reduction, and the capacity to provide both types of power compensation, it has a number of benefits.

Diesel fuel produces an average vibration level that is 21.1 m per second higher than any other fuel. On the other side, 16.8 m/s2 is the lowest average vibration level that can be achieved while using B20 test fuel and rotating at 2400 revolutions per minute. When traveling at the same speed, the B20 generates significantly different noise levels depending on whatever fuel is being used. Another thing that has been noticed is that the engine noise lowers above 2100 revolutions per minute for each kind of test fuel.

Furthermore, this research opens avenues for future studies focused on optimizing biodiesel blends with other renewable sources, exploring advanced fuel formulations, and investigating long-term impacts on engine wear and maintenance. By establishing a foundation for further exploration, this work contributes to a growing body of knowledge that can inform industry practices and guide future innovations in sustainable fuel technologies.In conclusion, the contributions of this study not only advance our understanding of biodiesel’s potential as a viable alternative fuel but also underscore its significance in addressing environmental challenges associated with conventional fossil fuels. As researchers and industry stakeholders continue to seek sustainable solutions for energy production, the insights gained from this research will be instrumental in shaping the future landscape of renewable energy utilization in internal combustion engines.