Master control loop of diesel generating set equipment

Abstract

The invention relates to the technical field of diesel generating sets, in particular to a master control loop of diesel generating set equipment. The system comprises a data acquisition module, a model creation module, a dynamic adjustment module and a multi-machine parallel module. Power change data of a load and rotating speed data of an engine are acquired through the data acquisition module, a long-short-term memory network model for predicting the rotating speed of the engine is constructed by utilizing the model creation module, and the fuel injection quantity of the engine is calculated by utilizing a PID control algorithm according to the predicted rotating speed value of the engine through the dynamic adjustment module. According to the change of load power, the fuel injection quantity of an engine is dynamically adjusted, the rotating speed of the engine is regulated and controlled to be in a stable state, it is guaranteed that a generator outputs stable voltage, a multi-machine parallel module controls the output power and frequency of each unit through a droop control method, the frequency is automatically adjusted, and phase embedding is synchronous; and multi-machine parallel loading is realized.

Description

Technical Field

[0001] The invention relates to the technical field of diesel generator sets, in particular to a general control loop of diesel generator set equipment. Background Art

[0002] The overall control loop of the diesel generator set is the core component of the diesel generator control system. It is responsible for monitoring and managing the operating status of the generator set to ensure that it can work stably and reliably under various working conditions. The overall control loop continuously improves and perfects the control system based on the power, load requirements, operating environment, etc. of the generator set to improve the performance and stability of the diesel generator and meet the needs of various application scenarios.

[0003] At present, when the load connected to a diesel generator set changes, the engine speed will change, which will in turn affect the output voltage and frequency of the generator and the power generation efficiency of the generator. In order to dynamically adjust the engine's fuel injection amount and regulate the engine speed to keep it in a stable state according to changes in load power, and ensure that the generator outputs a stable voltage, at the same time, when the load power increases the rated power of the unit, it can control subsequent units, automatically adjust the frequency and phase embedded synchronization, and realize multi-machine parallel loading. Therefore, we propose a total control loop for diesel generator set equipment. Summary of the invention

[0004] The purpose of the present invention is to solve the problem that when the load connected to the diesel generator set changes, the engine speed will change, thereby affecting the output voltage and frequency of the generator. In order to be able to dynamically adjust the engine's fuel injection amount and regulate the engine speed to keep it in a stable state according to the change of load power, to ensure that the generator outputs a stable voltage, at the same time, when the load power increases the rated power of the unit, it can control the subsequent units, automatically adjust the frequency and phase embedded synchronization, and realize multi-machine parallel loading.

[0005] To achieve the above object, the present invention provides a general control loop of a diesel generator set, comprising a data acquisition module, a model creation module, a dynamic adjustment module and a multi-machine parallel module; The data acquisition module collects the power change data of the load and the speed data of the engine, cleans the collected data, removes abnormal values ​​and erroneous data points, uses Z-Score normalization to normalize the load power and engine speed data, and transmits the processed data to the model creation module; The model creation module receives the power data and engine speed data collected by the data acquisition module, takes the load power data as input and the predicted engine speed as output, builds a long short-term memory network model for predicting the engine speed, divides the preprocessed data set into a training set, a validation set and a test set, uses a mean square error loss function to measure the difference between the predicted speed and the actual speed, trains the model, substitutes the power data into the model, obtains the predicted engine speed value, and transmits it to the dynamic adjustment module; The dynamic adjustment module compares the predicted engine speed value obtained by the model creation module with the rated engine speed value, calculates the fuel injection amount of the engine using the PID control algorithm, and converts it into a fuel injection amount adjustment signal, adjusts the output voltage through the voltage regulation circuit, and controls the fuel injection amount of the engine; When the load power reaches the rated power of the unit, the multi-machine parallel module starts the subsequent units, uses the droop control method to control the power and frequency output by each unit, and adjusts the phase of the unit through the synchronization detection device. When the frequency and phase are adjusted to meet the parallel conditions, the control system will perform synchronization confirmation and issue a closing command to allow the subsequent units to join the parallel operation.

[0006] As a further improvement of the present technical solution, the model creation module includes a model building unit and a model training unit; The model building unit uses an integer multiple of the sensor acquisition frequency as a step length, load power data as input, and predicted engine speed as output, determines three long short-term memory network layers, and builds a long short-term memory network model for predicting engine speed; The model training unit adopts the Adam optimizer, and the mean square error loss function is used to measure the difference between the predicted speed and the actual speed to train the model.

[0007] As a further improvement of the technical solution, the model building unit adds a fully connected layer after the long short-term memory network layer, and the number of neurons is less than the number of neurons in the long short-term memory network layer.

[0008] As a further improvement of the present technical solution, the formula of the mean square error loss function adopted by the model training unit is: ; in, is the mean square error, is the total number of data points, To predict the speed, is the actual speed.

[0009] As a further improvement of the present technical solution, the model training unit uses grid search to adjust the hyperparameters of the long short-term memory network model.

[0010] As a further improvement of the technical solution, the dynamic adjustment module includes a control algorithm unit and a dynamic adjustment unit; The control algorithm unit compares the predicted engine speed value with the rated engine speed value to obtain the speed deviation, and uses the PID control algorithm to calculate the proportional control amount, the integral control amount and the differential control amount to calculate the fuel injection amount of the engine; The dynamic adjustment unit establishes a mathematical model between the fuel injection amount and the control signal according to the injector characteristic curve, adjusts the output voltage through the voltage regulation circuit, and controls the fuel injection amount of the engine.

[0011] As a further improvement of the present technical solution, the control algorithm unit determines the upper and lower limits of the engine fuel injection amount according to the theoretical combustion model of the engine, and limits the engine fuel injection amount calculated by the PID control algorithm.

[0012] As a further improvement of the technical solution, the dynamic adjustment unit includes a voltage regulation circuit, wherein the voltage regulation circuit includes a voltage regulator LM, a voltage regulator diode VD and an inductor L; Pin 1 of the voltage regulator LM is connected to the input voltage UI and to one end of the capacitor C1, pin 4 of the voltage regulator LM is connected to one end of the resistor R1 and to one end of the resistor R2, the other end of the resistor R1 is connected to the other end of the capacitor C1 and to pins 3 and 5 of the voltage regulator LM, pin 2 of the voltage regulator LM is connected to the cathode of the voltage regulator diode VD and to one end of the inductor L, the other end of the inductor L is connected to one end of the capacitor C3 and to the other end of the resistor R2, the capacitor C2 is connected to the resistor R2, and the capacitor C3 is connected to the output voltage UO.

[0013] As a further improvement of the technical solution, the formula of the droop control method adopted by the multi-machine parallel module is: ; in, is the output active power, is the initial active power, is the droop coefficient, is the actual frequency, is the rated frequency.

[0014] As a further improvement of the technical solution, after the multi-machine parallel module is closed, the subsequent units gradually increase the load by means of soft loading.

[0015] Compared with the prior art, the present invention has the following beneficial effects: 1. The overall control loop of the diesel generator set equipment collects the load power change data and the engine speed data through the data acquisition module, and normalizes them. The model creation module is used to build a long short-term memory network model for predicting the engine speed. The engine speed is predicted according to the detected load power. The dynamic adjustment module calculates the engine's fuel injection amount according to the predicted engine speed value using the PID control algorithm, and adjusts the output voltage through the voltage regulation circuit to control the engine's fuel injection amount, so as to dynamically adjust the engine's fuel injection amount according to the change of load power, regulate the engine speed, make it in a stable state, and ensure that the generator outputs a stable voltage; 2. The multi-machine parallel module uses the droop control method to control the power and frequency output by each unit, and adjusts the phase of the unit through the synchronization detection device. When the frequency and phase are adjusted to meet the parallel conditions, the control system will confirm the synchronization and issue a closing command to allow subsequent units to join the parallel operation. When the load power increases the rated power of the unit, it can control the subsequent units, automatically adjust the frequency and phase to embed synchronization, and realize multi-machine parallel operation. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 It is a schematic diagram of the overall process of the present invention; Figure 2 It is a schematic diagram of the overall details of the present invention; Figure 3 This is a voltage regulation circuit diagram of the present invention.

[0017] The meaning of each number in the figure is: 100. Data acquisition module; 200. Model creation module; 210. Model building unit; 220. Model training unit; 300. Dynamic adjustment module; 310. Control algorithm unit; 320. Dynamic adjustment unit; 400. Multi-machine parallel module. DETAILED DESCRIPTION

[0018] The following will be combined with the accompanying drawings in 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.

[0019] At present, when the load connected to the diesel generator set changes, the engine speed will change, which will in turn affect the output voltage and frequency of the generator. In order to dynamically adjust the engine's fuel injection amount and control the engine speed according to the change in load power to keep it in a stable state and ensure that the generator outputs a stable voltage, at the same time, when the load power increases the rated power of the unit, it can control the subsequent units, automatically adjust the frequency and phase embedded synchronization, and realize multi-machine parallel loading.

[0020] Therefore, the present invention proposes to collect the power change data of the load and the speed data of the engine through the data acquisition module, construct a long short-term memory network model for predicting the engine speed by using the model creation module, calculate the engine's fuel injection amount according to the predicted engine speed value by using the PID control algorithm in the dynamic adjustment module, and control the power and frequency output by each unit by using the droop control method in the multi-machine parallel module.

[0021] The details are as follows: See also Figure 1 As shown, the present invention provides a general control loop of a diesel generator set, including a data acquisition module 100, a model creation module 200, a dynamic adjustment module 300 and a multi-machine parallel module 400; The data acquisition module 100 collects the power change data of the load and the speed data of the engine, cleans the collected data, removes abnormal values ​​and erroneous data points, uses Z-Score normalization to normalize the load power and engine speed data, and transmits the processed data to the model creation module 200; The model creation module 200 receives the power data and engine speed data collected by the data collection module 100, takes the load power data as input and the predicted engine speed as output, builds a long short-term memory network model for predicting the engine speed, divides the preprocessed data set into a training set, a validation set and a test set, uses a mean square error loss function to measure the difference between the predicted speed and the actual speed, trains the model, substitutes the power data into the model, obtains the predicted engine speed value, and transmits it to the dynamic adjustment module 300; The dynamic adjustment module 300 compares the predicted engine speed value obtained by the model creation module 200 with the rated engine speed value, calculates the fuel injection amount of the engine using the PID control algorithm, and converts it into a fuel injection amount adjustment signal, and adjusts the output voltage through the voltage regulation circuit to control the fuel injection amount of the engine; When the load power reaches the rated power of the unit, the multi-machine parallel module 400 starts the subsequent units, uses the droop control method to control the power and frequency output by each unit, and adjusts the phase of the unit through the synchronization detection device. When the frequency and phase are adjusted to meet the parallel conditions, the control system will perform synchronization confirmation and issue a closing command to allow the subsequent units to join the parallel operation.

[0022] like Figure 2 As shown, the model creation module 200 includes a model building unit 210 and a model training unit 220; The model building unit 210 uses an integer multiple of the sensor acquisition frequency as a step size, load power data as input, and predicted engine speed as output, determines three long short-term memory network layers, and constructs a long short-term memory network model for predicting engine speed; The model training unit 220 uses an Adam optimizer, and a mean square error loss function is used to measure the difference between the predicted speed and the actual speed to train the model; A high-precision load sensor and engine speed sensor, as well as a data acquisition card, were installed on a diesel generator set to collect load active power and engine speed data at a frequency of 10 times per second. The data was collected for a week, and more than 600,000 sets of data were obtained. The collected data were cleaned to remove abnormal values ​​and erroneous data points. For example, when a load sensor fails and causes the load data to suddenly jump to a maximum or minimum value, these data points are deleted. The data is then normalized, and the load and speed data are mapped to the range of 0 to 1 to facilitate the training of the neural network; Taking integer multiples of the sensor acquisition frequency as the step size, load power data as input, and predicted engine speed as output, three long short-term memory network layers are determined to build a long short-term memory network (LSTM) model for predicting engine speed. The first layer of LSTM begins to learn and model the input load power time series. It will automatically capture the long-term dependency in load power changes based on the sequence information of the input data. The output of the previous layer of LSTM is used as the input of the next layer of LSTM. Each layer can learn time series patterns at different levels.

[0023] In order to better integrate features of the LSTM layer, the model building unit 210 adds a fully connected layer after the long short-term memory network layer, and the number of neurons is less than that of the long short-term memory network layer; The fully connected layer is used to integrate and transform the features output by the LSTM layer. It maps the output vector of the LSTM layer to a new vector space for subsequent prediction tasks. Each neuron in the fully connected layer is connected to all output neurons of the previous layer (i.e., the LSTM layer), which enables it to receive all feature information from the LSTM layer. During the connection process, each connection corresponds to a weight parameter, and these weights form a weight matrix. By adjusting these weights, the fully connected layer can give different importance to different features output by the LSTM layer, thereby achieving feature integration; The fully connected layer performs a linear transformation on the input features, that is, multiplying the input feature vector by the weight matrix and adding a bias vector to obtain a new output vector. This linear transformation can map the high-dimensional features output by the LSTM layer to a different vector space, realizing the transformation of feature dimensions.

[0024] In order to better measure the difference between the predicted speed and the actual speed, the formula of the mean square error loss function used by the model training unit 220 is: ; in, is the mean square error, is the total number of data points, To predict the speed, is the actual speed.

[0025] The value of MSE represents the average of the squares of the errors between the predicted speed and the actual speed. Since the errors are squared, MSE will give a greater penalty to larger errors, which makes the model pay more attention to samples with larger errors, thereby prompting the model to minimize these larger errors.

[0026] When comparing multiple different engine speed prediction models, select the model with the smallest MSE value as the optimal model. For example, you may compare the prediction effects of a simple linear regression model and a complex neural network model on engine speed. By calculating their MSE values ​​on the same test data set, you find that the MSE value of the neural network model is smaller, so you can think that the neural network model has better prediction ability on this problem; In the process of training the model, MSE can be used as a loss function to optimize the parameters of the model. For example, by using an optimization algorithm such as the gradient descent algorithm, the parameters of the model are continuously adjusted to gradually reduce the MSE. In each iteration, the MSE of the current predicted speed and the actual speed is calculated, and the gradient of the model parameters is updated according to the MSE until the MSE reaches a satisfactory smaller value or reaches the preset number of iterations.

[0027] In order to better adjust the hyperparameters of the long short-term memory network model, the model training unit 220 uses grid search to adjust the hyperparameters of the long short-term memory network model; Use the determined hyperparameter value range and step size to generate all possible hyperparameter combinations. This can be achieved through nested loops, where each hyperparameter is sequentially assigned a value within its value range according to the step size to form a multi-dimensional hyperparameter combination grid. All generated hyperparameter combinations are recorded for subsequent model training and evaluation. For each set of hyperparameter combinations, the model is trained and validated using the same training and validation data sets. During the training process, the performance index of the model on the validation data set, i.e., the mean square error (MSE), is recorded. Among all hyperparameter combinations, the one with the best performance index is selected as the optimal hyperparameter combination. The hyperparameter combination with the smallest MSE value is selected, and the model is retrained using the optimal hyperparameter combination and validated on an independent test data set to ensure the generalization ability of the model. If the performance of the model on the test set also meets the requirements, then the model can be applied to the actual engine speed prediction task.

[0028] The dynamic adjustment module 300 includes a control algorithm unit 310 and a dynamic adjustment unit 320; The control algorithm unit 310 compares the predicted engine speed value with the rated engine speed value to obtain the speed deviation, and uses the PID control algorithm to calculate the proportional control amount, the integral control amount and the differential control amount to calculate the fuel injection amount of the engine; The dynamic adjustment unit 320 establishes a mathematical model between the fuel injection amount and the control signal according to the injector characteristic curve, adjusts the output voltage through the voltage regulation circuit, and controls the fuel injection amount of the engine; Set the initial values ​​of the proportional coefficient (Kp), integral coefficient (Ki) and differential coefficient (Kd). Generally speaking, the value range of Kp is between 0.1 and 10, the value range of Ki is between 0.01 and 1, and the value range of Kd is between 0.01 and 0.5.

[0029] Among them, the proportional control amount is calculated according to the proportional coefficient Kp and the current deviation e(t). The role of proportional control is to make the control amount proportional to the deviation. The larger the deviation, the larger the adjustment range of the control amount, so as to quickly reduce the deviation. The deviation is integrated to obtain the integral term, and then multiplied by the integral coefficient Ki to obtain the integral control amount. The role of integral control is to eliminate the steady-state error of the system. When the deviation exists, the integral term will continue to accumulate, so that the control amount is continuously adjusted until the deviation is zero. The differential of the deviation is calculated and then multiplied by the differential coefficient Kd to obtain the differential control amount. The role of differential control is to adjust the control amount in advance according to the rate of change of the deviation. When the deviation changes rapidly, the differential control amount will produce a larger adjustment to suppress the rapid change of the deviation and improve the stability and dynamic response performance of the system.

[0030] In order to better limit the engine fuel injection amount, the control algorithm unit 310 determines the upper and lower limits of the engine fuel injection amount according to the theoretical combustion model of the engine, and limits the engine fuel injection amount calculated by the PID control algorithm; According to the engine's intake system design and the maximum intake capacity of the air filter, determine the maximum air flow that the engine can inhale at different speeds and loads. Combined with the air-fuel ratio requirements in the theoretical combustion model, calculate the maximum fuel injection amount required at the maximum air flow rate as an important basis for the upper limit of the fuel injection amount; The engine needs a certain amount of fuel to maintain a stable combustion process. According to the theoretical combustion model, the minimum injection amount required to ensure stable combustion of the engine under different working conditions is analyzed. If the injection amount is lower than this, the engine may experience unstable phenomena such as misfire and jitter. During engine operation, the PID control algorithm calculates the required fuel injection amount in real time based on the actual engine speed, load and other parameters. At the same time, it monitors the engine operating conditions in real time, including air flow, engine speed, load and other parameters, and substitutes these parameters into the theoretical combustion model to calculate the upper and lower limits of the fuel injection amount under the current operating conditions. The fuel injection amount calculated by the PID control algorithm is compared with the calculated upper and lower limits. If the fuel injection amount exceeds the upper limit, the fuel injection amount is limited to the upper limit value to avoid engine failure or performance degradation due to excessive fuel injection. If the fuel injection amount is lower than the lower limit, the fuel injection amount is adjusted to the lower limit value to ensure stable combustion of the engine and meet emission regulations.

[0031] like Figure 3 As shown, in order to better adjust the voltage of the fuel injector, the dynamic adjustment unit 320 includes a voltage regulation circuit, wherein the voltage regulation circuit includes a voltage regulator LM, a voltage regulator diode VD and an inductor L; Pin 1 of the voltage regulator LM is connected to the input voltage UI and to one end of the capacitor C1, pin 4 of the voltage regulator LM is connected to one end of the resistor R1 and to one end of the sliding variable resistor R2, the other end of the resistor R1 is connected to the other end of the capacitor C1 and to pins 3 and 5 of the voltage regulator LM, pin 2 of the voltage regulator LM is connected to the cathode of the voltage zener diode VD and to one end of the inductor L, the other end of the inductor L is connected to one end of the capacitor C3 and to the other end of the sliding variable resistor R2, the capacitor C2 is connected to the sliding variable resistor R2, and the capacitor C3 is connected to the output voltage UO; In this circuit, the resistor R1 and the sliding rheostat R2 form a voltage-dividing resistor network, which cooperates with the internal circuit of the voltage regulator LM to set the parameters such as the voltage-dividing ratio of the output voltage, thereby affecting the final output voltage UO. The performance of the converter is regulated by the inductor L. The larger the inductance value, the smaller the ripple of the output voltage, but the response speed of the circuit will become slower. The smaller the inductance value, the larger the ripple, but the response speed will become faster. The capacitor C1 is used to filter out the low-frequency noise and fluctuation in the input voltage, and the capacitor C3 is used to filter out the high-frequency noise and ripple in the output voltage, thereby improving the stability of the output voltage. According to the voltage information fed back by the voltage-dividing resistor network formed by the resistor R1 and the sliding rheostat R2, the output duty cycle is automatically adjusted, thereby adjusting the output voltage; For example, resistor R1 is , the sliding resistor R2 is , capacitor C1 is , capacitor C2 , capacitor C3 is , the inductance L is , the internal reference voltage of the regulator LM is ; The output voltage calculation formula is: ; in, is the output voltage, is the internal reference voltage of the regulator LM, is the resistance value of resistor R1, is the resistance of the sliding resistor R2.

[0032] When the resistance of the sliding resistor R2 is adjusted to the minimum When , the output voltage formula can be obtained: ; When the resistance of the sliding resistor R2 is adjusted to the maximum When , the output voltage formula can be obtained: ; By adjusting the resistance of the sliding resistor R2, the output voltage is guaranteed to be to It can be adjusted continuously to further adjust the engine's fuel injection amount.

[0033] In order to better control the power and frequency output by each unit, the formula of the droop control method adopted by the multi-machine parallel module 400 is: ; in, is the output active power, is the initial active power, is the droop coefficient, is the actual frequency, is the rated frequency.

[0034] Droop control is based on the active power-frequency droop characteristics of the generators in the power system. Ideally, the frequency of the power system is constant, but when the active power demand in the system changes, the frequency will fluctuate accordingly. Droop control simulates the natural droop characteristics of the generator, that is, there is an approximately linear inverse relationship between active power and frequency, to adjust the output power of each unit to maintain the stability of the system frequency. In a multi-unit system, droop control must not only achieve frequency stability, but also ensure the reasonable distribution of active power among the units. A power distribution strategy is formulated based on factors such as the rated capacity of each unit and the droop coefficient. For example, the principle of equal proportion distribution is adopted, that is, each unit shares the change in active power according to the proportion of its rated capacity to the total capacity of the system. When the system frequency drops, each unit increases its output power according to the droop characteristics, and the increased power is distributed according to the power distribution strategy to avoid the situation where some units are overloaded while other units are not fully utilized. In order to achieve coordinated control of multiple units, information communication is required between the units. By establishing a communication network, such as a communication bus or Ethernet dedicated to the power system, each unit sends its own operating status information, such as power, frequency, droop coefficient, etc. to other units and the system control center. The system control center coordinates and optimizes according to the global information, adjusts the droop coefficient of each unit or sends control instructions to achieve optimal operation of the entire power system.

[0035] In order to better add the subsequent units to the parallel operation, after the multi-machine parallel module 400 is closed, the subsequent units gradually increase the load by soft loading; Use the segmented loading in soft loading to control the subsequent units to join the parallel operation. Before starting loading, let the system or equipment run for a period of time in the no-load state to check whether its various performances are normal, such as whether the equipment starts smoothly and whether the control system responds sensitively. After the no-load operation is normal, enter the low-load operation stage and increase the load to about 20% to 30% of the rated load. Observe the operation of the system under low load, including the vibration, noise, heat generation of the equipment, and the control accuracy and stability of the system. After the low-load operation is stable, gradually increase the load to the medium load and high load stages. The magnitude of each load increase can be determined according to the system's tolerance and actual needs; Analyze the load changes of the system or equipment during actual use to obtain its typical load curve. For example, for a power system, analyze its load change rules at different times of the day, including peak hours, valley hours and average load, etc. According to the actual load curve obtained, simulate the curve for load loading during the test or operation. This can be achieved by writing a loading program or using a load device with curve simulation function.

[0036] In summary, the working principle of this solution is as follows: The overall control loop of the diesel generator set equipment collects the power change data of the load and the speed data of the engine through the data acquisition module 100, and normalizes them. The model creation module 200 is used to build a long short-term memory network model for predicting the engine speed, and the engine speed is predicted according to the detected load power. The dynamic adjustment module 300 calculates the engine injection amount according to the predicted engine speed value using the PID control algorithm, and adjusts the output voltage through the voltage regulation circuit to control the engine injection amount, so as to dynamically adjust the engine injection amount according to the change of the load power, regulate the engine speed, make it in a stable state, and ensure that the generator outputs a stable voltage. The multi-machine parallel module 400 uses the droop control method to control the power and frequency output by each unit, and adjusts the phase of the unit through the synchronization detection device. When the frequency and phase are adjusted to meet the parallel condition, the control system will perform synchronization confirmation and issue a closing command to allow the subsequent units to join the parallel operation. When the load power increases the rated power of the unit, it can control the subsequent units, automatically adjust the frequency and phase to embed synchronization, and realize multi-machine parallel operation with load.

[0037] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited by the above embodiments. The above embodiments and descriptions are only preferred examples of the present invention and are not intended to limit the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, which fall within the scope of the present invention. The scope of protection of the present invention is defined by the attached claims and their equivalents.

Claims

1. A general control circuit of a diesel generator set, characterized in that: It comprises a data collection module (100), a model creation module (200), a dynamic adjustment module (300) and a multi-machine parallel module (400); The data acquisition module (100) acquires load power variation data and engine speed data, cleans the acquired data, removes abnormal values ​​and erroneous data points, normalizes the load power and engine speed data using Z-Score normalization, and transmits the processed data to the model creation module (200); The model creation module (200) receives the power data and engine speed data collected by the data collection module (100), takes the load power data as input and the predicted engine speed as output, constructs a long short-term memory network model for predicting the engine speed, divides the preprocessed data set into a training set, a validation set and a test set, uses a mean square error loss function to measure the difference between the predicted speed and the actual speed, trains the model, substitutes the power data into the model, obtains the predicted engine speed value, and transmits it to the dynamic adjustment module (300); The dynamic adjustment module (300) compares the predicted engine speed value obtained by the model creation module (200) with the rated engine speed value, calculates the fuel injection amount of the engine using a PID control algorithm, converts the calculated fuel injection amount into a fuel injection amount adjustment signal, and adjusts the output voltage through a voltage adjustment circuit to control the fuel injection amount of the engine; When the load power reaches the rated power of the unit, the multi-machine parallel module (400) starts the subsequent units, controls the power and frequency output by each unit using a droop control method, and adjusts the phase of the units using a synchronization detection device. When the frequency and phase are adjusted to meet the parallel operation conditions, the control system performs synchronization confirmation and issues a closing command to allow the subsequent units to join the parallel operation.

2. The overall control circuit of the diesel generator set according to claim 1 is characterized in that: The model creation module (200) includes a model building unit (210) and a model training unit (220); The model building unit (210) uses an integer multiple of the sensor acquisition frequency as a step length, load power data as input, and predicted engine speed as output, determines three long short-term memory network layers, and builds a long short-term memory network model for predicting engine speed; The model training unit (220) uses an Adam optimizer, and a mean square error loss function is used to measure the difference between the predicted rotation speed and the actual rotation speed to train the model.

3. The overall control circuit of the diesel generator set according to claim 2 is characterized in that: The model building unit (210) adds a fully connected layer after the long short-term memory network layer, and the number of neurons is less than the number of neurons in the long short-term memory network layer.

4. The overall control circuit of the diesel generator set according to claim 2 is characterized in that: The formula of the mean square error loss function used by the model training unit (220) is: ； in, is the mean square error, is the total number of data points, To predict the speed, is the actual speed.

5. The overall control circuit of the diesel generator set according to claim 4 is characterized in that: The model training unit (220) uses grid search to adjust the hyperparameters of the long short-term memory network model.

6. The overall control circuit of the diesel generator set according to claim 1 is characterized in that: The dynamic adjustment module (300) comprises a control algorithm unit (310) and a dynamic adjustment unit (320); The control algorithm unit (310) compares the predicted engine speed value with the rated engine speed value to obtain a speed deviation, calculates a proportional control amount, an integral control amount and a differential control amount using a PID control algorithm, and calculates a fuel injection amount of the engine; The dynamic adjustment unit (320) establishes a mathematical model between the fuel injection quantity and the control signal according to the fuel injector characteristic curve, and adjusts the output voltage through the voltage adjustment circuit to control the fuel injection quantity of the engine.

7. The overall control circuit of the diesel generator set according to claim 6 is characterized in that: The control algorithm unit (310) determines the upper and lower limits of the engine fuel injection amount according to the theoretical combustion model of the engine, and limits the engine fuel injection amount calculated by the PID control algorithm.

8. The overall control circuit of the diesel generator set according to claim 6 is characterized in that: The dynamic adjustment unit (320) comprises a voltage regulation circuit, wherein the voltage regulation circuit comprises a voltage regulator LM, a voltage regulator diode VD and an inductor L; Pin 1 of the voltage regulator LM is connected to the input voltage UI and to one end of the capacitor C1, pin 4 of the voltage regulator LM is connected to one end of the resistor R1 and to one end of the resistor R2, the other end of the resistor R1 is connected to the other end of the capacitor C1 and to pins 3 and 5 of the voltage regulator LM, pin 2 of the voltage regulator LM is connected to the cathode of the voltage regulator diode VD and to one end of the inductor L, the other end of the inductor L is connected to one end of the capacitor C3 and to the other end of the resistor R2, the capacitor C2 is connected to the resistor R2, and the capacitor C3 is connected to the output voltage UO.

9. The overall control circuit of the diesel generator set according to claim 1 is characterized in that: The formula of the droop control method adopted by the multi-machine parallel module (400) is: ； in, is the output active power, is the initial active power, is the droop coefficient, is the actual frequency, is the rated frequency.

10. The overall control circuit of the diesel generator set according to claim 9 is characterized in that: After the multi-machine parallel module (400) is switched on, the subsequent units gradually increase the load by means of soft loading.

Images