A brushless generator with overheat alarm and early warning method thereof

By incorporating an overheat alarm device and intelligent algorithms within the brushless generator, the alarm problem when the brushless generator overheats is solved, enabling timely detection and resolution of overheating issues, extending equipment lifespan, and improving generator operating efficiency and reliability.

CN117713457BActive Publication Date: 2024-11-01JIANGSU HEISONG TECH CO LTD
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202311609967.5
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2023-11-29
Publication Date
2024-11-01
Estimated Expiration
2043-11-29

AI Technical Summary

Technical Problem

Existing brushless generators cannot provide timely alarms when overheating, leading to damage to mechanical components and affecting equipment lifespan.

Method used

An overheat alarm device is installed inside the brushless generator. The alarm switch is triggered by the expansion of kerosene, and the temperature change trend is predicted by the gear transmission system and intelligent algorithm. The working state is optimized by combining machine learning algorithm.

Benefits of technology

It enables timely detection and resolution of generator overheating issues, extends equipment lifespan, and improves generator operating efficiency and reliability.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN117713457B_ABST
    Figure CN117713457B_ABST
Patent Text Reader

Abstract

The present invention provides a brushless generator with overheat alarm and an early warning method thereof, and relates to the field of overheat alarm. The brushless generator with overheat alarm comprises a generator body, the top of the generator body is fixedly connected with an alarm mechanism, the touch switch of the alarm mechanism is located inside the generator body, an overheat alarm device is arranged inside the generator body, the overheat alarm device contains an overheat alarm control device, the overheat alarm control device is fixedly connected to the inner wall of the generator body, the overheat alarm device is fixedly connected to the inner wall of the generator body, and the overheat alarm device is located at the bottom of the touch switch of the alarm mechanism. The kerosene expands due to heat, causing the piston to move upward, so that the touch plate touches the switch of the alarm mechanism, does not require any external power supply, works stably and reliably, can work normally in various harsh environments, is not easily affected by factors such as electromagnetic interference, and the alarm signal sent is reliable.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of overheat alarm, and in particular to a brushless generator with overheat alarm and an early warning method thereof. Background Art

[0002] The brushless generator is a new type of generator based on electronic control technology, also called a permanent magnet synchronous generator, which consists of permanent magnets, rotors, stators, end covers, etc. During operation, the magnetic field generated by the permanent magnets will act on the rotor, causing the rotor to start rotating and generating a three-phase potential difference on the stator. These potential differences are then output as stable alternating current through the electronic controller. At the same time, the electronic controller will also control the motor's speed and power output and other parameters to ensure its normal operation. However, it consumes a certain amount of energy when working, and also generates a lot of heat energy. If the heat cannot be dissipated in time, it will cause the generator to overheat or even fail, affecting the normal operation of the generator. Therefore, an overheating alarm device is urgently needed to sound an alarm when the generator temperature exceeds the set warning value to remind the operator to handle it.

[0003] A Chinese invention patent with publication number CN109194034A discloses a high-efficiency brushless generator, which fixes the rotating shaft and inserts a limit ring into an annular groove. The limit rod plays a blocking role to prevent the rotating shaft from sliding along the axial direction of the shell. An annular groove is used, so that even when the rotating shaft rotates, the rotating shaft can be limited to ensure the fixing effect of the rotating shaft. However, it cannot sound an alarm in the event of overload or high temperature. Excessive temperature will cause the materials of components such as the rotor and stator to deform or age, accelerate the wear and aging of the equipment, and thus shorten its service life. Summary of the invention

[0004] 1. Technical issues to be solved

[0005] In view of the deficiencies in the prior art, the present invention provides a brushless generator with overheat alarm and an early warning method thereof, which solves the problem that when the temperature of the existing brushless generator is too high, the internal mechanical components of the generator are damaged.

[0006] (II) Technical solution

[0007] To achieve the above objectives, the present invention is implemented through the following technical solutions: a brushless generator with overheating alarm, comprising a generator body, an alarm mechanism fixedly connected to the top of the generator body, a touch switch of the alarm mechanism located inside the generator body, an overheating alarm device provided inside the generator body, the overheating alarm device containing an overheating alarm control device, the overheating alarm control device fixedly connected to the inner wall of the generator body, the overheating alarm device fixedly connected to the inner wall of the generator body, and the overheating alarm device located at the bottom of the touch switch of the alarm mechanism.

[0008] The above technical solution can help timely discover and solve the overheating problem of the brushless generator, thereby ensuring the normal operation of the generator and extending the life of the machine.

[0009] Preferably, the overheating alarm device includes a flat plate, both ends of which are fixedly connected to the inside of the generator body, a plurality of T-shaped support plates are fixedly connected to the top surface of the flat plate, a rotating rod is rotatably connected between the plurality of T-shaped support plates, the rotating rod passes through both sides of the T-shaped support plates, and the through ends on both sides of the rotating rod are respectively fixedly connected to the first gear.

[0010] Preferably, first support columns are provided on both sides of the top of the flat plate, and the first support columns are hollow cylindrical structures. The first support columns on one side are through-connected with the flat plate, and a telescopic column is slidably connected inside the first support column. A plurality of teeth are provided on the surface of the telescopic column facing the first gear and the second gear, and an arc-shaped groove is provided on the surface of the first support column, and the arc-shaped groove is located directly opposite to the first gear and the second gear. The first gear and the second gear are located in the arc-shaped groove and mesh with the teeth on the surface of the telescopic column to rotate.

[0011] Through the above technical solution, the kerosene expansion can trigger the alarm switch. Once the generator stops working or fails for some reason, it can be detected and an alarm signal can be issued in the first time.

[0012] Preferably, the top of the telescopic column is fixedly connected with a touch plate, and the touch plate is located at the bottom of the alarm mechanism, and the bottom of the flat plate is fixedly connected with a heated cylinder, and the heated cylinder is located at the bottom of the first support column which is connected to the first gear transmission, and the heated cylinder is through-connected with the first support column, and kerosene is provided inside the heated cylinder, and the bottom end of the telescopic column is fixedly connected with a piston, and the piston is slidably connected to the inside of the heated cylinder, and a first connecting block is fixedly connected with the center position of the top surface of the flat plate, the first connecting block is rotatably connected with the first connecting plate, the end of the first connecting plate away from the first connecting block is rotatably connected with the second connecting plate, and the end of the second connecting plate away from the first connecting plate is rotatably connected with the first connecting block, and the first connecting block is fixedly connected to the bottom surface of the touch plate.

[0013] Through the above technical solution, the kerosene expansion trigger alarm switch does not require the support of an external power supply and controller, is relatively simple to use, does not require frequent maintenance and servicing, and does not require replacement of batteries and other components, making maintenance and management of the equipment more convenient.

[0014] Preferably, a pressing disc is fixedly connected to the top of the pressing column, the bottom end of the first spring is fixedly connected to the bottom end of the pressing disc, a slider is slidably connected in the Z-shaped slide groove of the movable column, the slider is fixedly connected to the support plate through a U-shaped connecting frame, and the slider is slidably connected to the U-shaped connecting frame, a metal strip is provided at one end of the slider away from the movable column, the metal strip is fixedly connected to the U-shaped connecting frame, the metal strip is made of elastic and recoverable metal, and a rotating plate is fixedly connected to the bottom end of the movable column.

[0015] Preferably, two limit columns are provided at the bottom of the rotating plate, and the two limit columns are fixedly connected to the support plate, a second support column is provided between the two limit columns, and the second support column is fixedly connected to the support plate, the second support column is rotatably connected to a rotating frame, and limit rods are fixedly connected on both sides of the connection between the rotating frame and the second support column, a sliding rod is fixedly connected to the bottom of the rotating frame, and the sliding rod is slidably connected to a rocker rod, a sliding groove is provided inside the rocker rod, and the sliding rod is located in the sliding groove inside the rocker rod.

[0016] Preferably, the outer surface of the rocker arm is slidably connected to a second spring, the bottom end of the second spring is fixedly connected to the outer surface of the rocker arm, the bottom end of the rocker arm is slidably connected to the limit frame, the outer surface of the rocker arm is rotatably connected to a push rod, and the push rod is rotatably connected to the first connecting plate.

[0017] Through the above technical solution, the touch panel can be separated from the alarm mechanism switch by moving the pressing disk, so that the user can manually cancel the alarm to avoid misjudgment or waste of time and resources due to false alarms.

[0018] The present invention provides an early warning method for a brushless generator with an overheat alarm, comprising the following specific steps:

[0019] S1. When the temperature inside the generator body (1) is too high, the kerosene (18) expands due to the heat, causing the telescopic column (8) to move upward;

[0020] S2, the teeth (11) on the telescopic column (8) drive the first gear (6) to rotate, and under the cooperation of the rotating rod (5) and the second gear (7), the telescopic columns (8) on both sides move upward synchronously, so that the touch plate (9) moves upward, presses the switch of the alarm mechanism (2), and the alarm mechanism (2) sends out an alarm message;

[0021] S3. Under the action of the first connecting plate (14) and the second connecting plate (15), the touch plate (9) can be kept in contact with the switch of the alarm mechanism (2), thereby preventing the touch plate (9) from falling, so that the staff misses the alarm information, helping to timely discover and solve the problem of overheating of the generator, preventing the occurrence of faults, protecting the generator, and extending its service life;

[0022] S4, when the alarm needs to be cancelled, the pressing disc (26) is moved to move the moving column (23) downward;

[0023] S5. Under the action of the Z-shaped slide groove (24) and the slider (29) inside the movable column (23), the movable column (23) rotates when moving downward, so that the rotating plate (31) can squeeze the limit rods (40) on both sides back and forth, without the need for plugging and unplugging;

[0024] S6, the disc (26) is pressed, the limit rod (40) moves, and the rotating frame (34) is caused to rotate. Under the action of the second spring (38), the swing rod (36) is caused to rotate left and right on the limit frame (39), driving the push rod (20) to deflect, causing the first connecting plate (14) and the second connecting plate (15) to rotate, so that the touch plate (9) moves downward, thereby canceling the alarm. The operation is simple and convenient for daily use.

[0025] As an embodiment of the present invention, it is also possible to use an intelligent algorithm to predict the change trend of the internal temperature of the generator body between steps S1 and S2:

[0026] 1) Monitor the temperature change of the generator in real time through the temperature sensor installed inside the generator body, and transmit the temperature data to the processor;

[0027] 2) Preprocess the temperature data and integrate the data from different sensors;

[0028] 3) Extract features related to temperature change trends from temperature data, including time series features and sliding window average features;

[0029] 4) Use regression model to train temperature data to learn the trend and pattern of temperature change. The specific application case of the detailed formula derivation process of this algorithm is as follows:

[0030] Collect a certain amount of historical temperature data as a training data set, and then use the least squares method to fit these data to obtain the relationship model between temperature and related factors:

[0031] Assume that there are n historical temperature data samples, where the temperature of the i-th sample is Ti, and the corresponding related factors are xi, including voltage, power, and load. The following equations are obtained using the least squares method:

[0032] ∑(T i-θ0-θ1xi)=0

[0033] ∑(xi-μ) 2 =σ 2

[0034] Among them, θ0 and θ1 are the parameters of the model, representing the intercept and slope respectively; μ and σ represent the mean and standard deviation of the independent variables; Ti and xi represent the actually observed temperature and the values ​​of related factors respectively;

[0035] By solving this set of equations, the parameters θ0 and θ1 of the model can be obtained, thus obtaining the relationship model between temperature and related factors;

[0036] The following is a prediction of the newly collected temperature data:

[0037] Assume there are n temperature samples, each containing m features and a label: X = [x_1, x_2, ..., x_m], representing the time feature, and y = [y_1, y_2, ..., y_n], representing the actual temperature. The goal of linear regression is to find a straight line y = w*x+b that minimizes the error between the predicted value y' and the true value y:

[0038] First, calculate the mean vectors X_bar and Y_bar of the samples:

[0039] X_bar=(1 / n)*∑(x_i)

[0040] Y_bar=(1 / n)*∑(y_i)

[0041] Then, calculate the covariance matrix S of the sample:

[0042] S=Σ((x_i-X_bar)(x_j-X_bar)^T)

[0043] Then calculate the transposed matrix X^T of the sample's independent variable X:

[0044] X^T=[x_1,x_2,...,x_m]^T

[0045] Finally, solve for the linear regression coefficient w and intercept b:

[0046] w=(X^T*S^-1*X)^-1*X^T*S^-1*Y

[0047] b=Y_bar-w*X_bar

[0048] After obtaining the linear regression model, for the new sample x, it can be predicted by the following formula:

[0049] y'=w*x+b

[0050] 5) Use the trained regression model to predict the newly collected temperature data and obtain the change trend of the internal temperature of the generator;

[0051] 6) As a double insurance, according to the relationship model between temperature and related factors, as well as the prediction results of the change trend of the internal temperature of the generator, it can be judged whether it is necessary to trigger the alarm mechanism to alarm. If the prediction results show that the temperature will exceed the safe range or there is a risk of overheating, the alarm mechanism will be triggered to sound an alarm;

[0052] After step S6, the working state of the generator is optimized by an intelligent algorithm to improve its efficiency and reliability. The specific implementation process of introducing the combination of two machine learning algorithms is as follows:

[0053] 1) Collect data on the operating status and environmental parameters of the generator, including real-time data collected by temperature, humidity, and pressure sensors, as well as the operating parameters of the generator, including voltage, current, and power;

[0054] 2) Based on the collected data, feature engineering processing is required to extract features that affect the working status and efficiency of the generator, including calculating the average, maximum and minimum statistical features of temperature and humidity, or obtaining new features through some transformation of sensor data;

[0055] 3) Select two different machine learning algorithms to train the model, use the decision tree algorithm for feature selection and classification, determine which features have the greatest impact on generator efficiency and reliability through analysis of historical data, and establish a decision tree model for classification, divide the load into two categories: high load and low load, and make judgments based on temperature and humidity characteristics; then use the neural network algorithm for parameter optimization and prediction, divide the data set into training set and test set according to the results of the decision tree model, use the training set to train the neural network model, and calculate the model parameters: weight and bias; use the test set to evaluate the model, calculate the prediction error, and select the optimal model to predict the best working parameters in the future;

[0056] In this step, a decision tree algorithm is used for feature selection and classification to determine which features have the greatest impact on generator efficiency and reliability, and a decision tree model is built to classify the load into two categories: high load and low load;

[0057] Feature selection: During the feature selection process, the importance of each feature is evaluated using Information Gain or Gini Impurity metrics, which are often used to determine the splitting point of the feature.

[0058] The formula for Input Gain is:

[0059] IG(D,A)=H(D)-∑ v∈Values(A) (∣D∣ / ∣Dv∣)H(Dv)

[0060] in:

[0061] IG(D,A) is the information gain of feature A;

[0062] H(D) is the entropy of the data set D, which is used to measure the uncertainty of the data;

[0063] Values(A) are the possible values ​​of feature A;

[0064] Dv is the data subset whose feature A value is v;

[0065] |D| represents the size of the data set D;

[0066] The formula for Gini impurity is:

[0067]

[0068] in:

[0069] G ini(D) is the Gini impurity of data set D;

[0070] C is the set of categories;

[0071] p(i|D) is the proportion of samples belonging to category i in the dataset D;

[0072] By calculating information gain or Gini impurity, we can determine which feature has the most discriminative power in the classification task;

[0073] Decision Tree Classification: Once the key features are selected, a decision tree algorithm can be used to build a classification model. The decision tree recursively selects the best features to split the data until the stopping condition is met. This will form a tree structure for classifying new data.

[0074] Next, we will use the neural network algorithm for parameter optimization and prediction. Based on the results of the decision tree model, we will divide the data set into training and test sets.

[0075] Split the dataset into training sets, which can be achieved using the following formula:

[0076] Training set: Dtra in = (Xtra in, Ytra in)

[0077] Test set: Dtest = (Xtest, Ytest)

[0078] Among them, Xtra in and Xtest are feature data, Ytra in and Ytest are corresponding labels or outputs;

[0079] When training a neural network model, an optimization algorithm such as gradient descent is usually used to calculate the model parameters: weights and biases. The goal of training is to minimize the loss function, and the root mean square error (RMSE) is usually used to measure the prediction error. The optimization objective function can be expressed as:

[0080]

[0081] Among them, J(θ) is the loss function, θ is the model parameter, m is the number of training samples, hθ(x (i) ) is the model's prediction for sample x (i) The prediction of (i) is the actual label;

[0082] Use the test set to evaluate the trained neural network model and calculate the prediction error metrics such as the root mean square error (RMSE). The formula for RMSE is:

[0083]

[0084] Where n is the number of test samples, y (i) test is the actual label, y^ (i) test is the predicted value of the model;

[0085] Predicting future working parameters: Select the optimal neural network model to predict the best working parameters in the future, input new feature data into the model, and obtain the prediction results. This can be achieved through the forward propagation process of the neural network;

[0086] 4) Finally, the working state of the generator is adjusted according to the prediction results, the calculated optimal working parameters are transmitted to the control system, and the load and speed parameters of the generator are adjusted to achieve the purpose of improving efficiency and reliability. At the same time, the working state and environmental parameters of the generator are monitored in real time and dynamically adjusted as needed. If any abnormal situation or inaccurate prediction is found, the data can be collected again and the model can be updated;

[0087] 5) During the operation of the generator, it is necessary to monitor the working status and environmental parameters of the generator in real time. Once an abnormal situation is detected or the environmental parameters change, the trained model can be used for dynamic adjustment to optimize the working status of the generator. This process can be repeated until the generator shuts down or reaches the preset stop condition.

[0088] The present invention provides a brushless generator with overheat alarm and an early warning method thereof. It has the following beneficial effects:

[0089] The present invention arranges an overheat alarm device inside the generator body, and the kerosene expands when heated, so that when the kerosene expands when heated, it drives the piston to move upward, so that the telescopic column moves upward synchronously, and with the cooperation of the meshing teeth and gears on the telescopic column, the telescopic columns on both sides drive the touch plates to move upward, and touch the switch of the alarm mechanism to sound an alarm, which can help to timely discover and solve the problem of overheating of the generator, prevent the occurrence of malfunctions, protect the generator, and extend the service life. The kerosene does not need the support of an external power supply and a controller when it expands when heated, so it is relatively simple to use and does not need frequent maintenance.

[0090] A regression algorithm is added between steps S1 and S2 to predict the changing trend of the temperature inside the generator body. By predicting the changing trend of the temperature inside the generator body, possible faults or abnormal conditions can be discovered in advance, so that timely measures can be taken for maintenance or adjustment to avoid accidents. According to the predicted temperature change trend, the working parameters of the generator, such as load, speed, etc., can be adjusted to optimize the operating state of the generator, reduce temperature risks, and improve the efficiency and reliability of the generator. After step S6, an intelligent algorithm is added to optimize the working state of the generator, which can improve its efficiency and reliability. By optimizing the working state of the generator, such as adjusting parameters such as load and speed, the power generation efficiency of the generator can be improved. By optimizing the working state, the load and wear of the generator can be reduced, the failure rate can be reduced, and the service life of the generator can be extended. By real-time monitoring and optimizing the working state of the generator, potential problems can be discovered in time and measures can be taken to solve them, thereby improving the reliability and stability of the generator. BRIEF DESCRIPTION OF THE DRAWINGS

[0091] Figure 1 It is a schematic diagram of the three-dimensional structure of the present invention;

[0092] Figure 2 This is a side overall structural diagram of the overheat alarm device of the present invention;

[0093] Figure 3 A cross-sectional view of a portion of the overheat alarm device of the present invention;

[0094] Figure 4 For the present invention Figure 3 The three-dimensional structure diagram of

[0095] Figure 5 It is a three-dimensional structural diagram of the overheat alarm control device of the present invention.

[0096] Among them, 1. generator body; 2. alarm mechanism; 3. plate; 4. T-shaped support plate; 5. rotating rod; 6. first gear; 7. second gear; 8. telescopic column; 9. touch plate; 10. first support column; 11. meshing teeth; 12. arc-shaped notch; 13. first connecting block; 14. first connecting plate; 15. second connecting plate; 16. first connecting block; 17. piston; 18. kerosene; 19. heating cylinder; 20. push rod; 21. Support plate; 22. Extension plate; 23. Moving column; 24. Z-shaped slide groove; 25. Pressing column; 26. Pressing disc; 27. First spring; 28. U-shaped connecting frame; 29. ​​Sliding block; 30. Metal strip; 31. Turntable; 32. Limiting column; 33. Second supporting column; 34. Rotating frame; 35. Sliding rod; 36. Rocker arm; 37. Slide groove; 38. Second spring; 39. Limiting frame; 40. Limiting rod. DETAILED DESCRIPTION

[0097] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0098] like Figure 1 and Figure 2 It is shown that an embodiment of the present invention provides a brushless generator with overheating alarm, including a generator body 1, an alarm mechanism 2 is fixedly connected to the top of the generator body 1, the touch switch of the alarm mechanism 2 is located inside the generator body 1, an overheating alarm device is provided inside the generator body 1, the overheating alarm device contains an overheating alarm control device, the overheating alarm control device is fixedly connected to the inner wall of the generator body 1, the overheating alarm device is fixedly connected to the inner wall of the generator body 1, and the overheating alarm device is located at the bottom of the touch switch of the alarm mechanism 2.

[0099] like Figure 3 and Figure 4 As shown, the overheat alarm device includes a flat plate 3, and first support columns 10 are arranged on both sides of the top of the flat plate 3. The first support column 10 is a hollow cylindrical structure, and the first support column 10 on one side is connected with the flat plate 3. A telescopic column 8 is slidably connected inside the first support column 10, and a touch plate 9 is fixedly connected to the top of the telescopic column 8. The touch plate 9 is located at the bottom of the alarm mechanism 2, and a heated cylinder 19 is fixedly connected to the bottom of the flat plate 3. The heated cylinder 19 is located at the bottom of the first support column 10 that is transmission-connected to the first gear 6, and the heated cylinder 19 is connected with the first support column 10.

[0100] Kerosene 18 is arranged inside the heating cylinder 19, a piston 17 is fixedly connected to the bottom end of the telescopic column 8, and the piston 17 is slidably connected to the inside of the heating cylinder 19. A first connecting block 13 is fixedly connected to the center position of the top surface of the flat plate 3, and the first connecting block 13 is rotatably connected to the first connecting plate 14, and the end of the first connecting plate 14 away from the first connecting block 13 is rotatably connected to the second connecting plate 15, and the end of the second connecting plate 15 away from the first connecting plate 14 is rotatably connected to the first connecting block 16, and the first connecting block 16 is fixedly connected to the bottom surface of the touch plate 9, and the telescopic column 8 faces the first gear 6 and the second gear A plurality of teeth 11 are provided on the surface of the wheel 7, an arcuate notch 12 is provided on the surface of the first support column 10, the arcuate notch 12 is located directly opposite to the first gear 6 and the second gear 7, the first gear 6 and the second gear 7 are located in the arcuate notch 12 and mesh with the teeth 11 on the surface of the telescopic column 8 for rotation, the two ends of the plate 3 are fixedly connected to the inside of the generator body 1, a plurality of T-shaped support plates 4 are fixedly connected to the top surface of the plate 3, a rotating rod 5 is rotatably connected between the plurality of T-shaped support plates 4, the rotating rod 5 passes through both sides of the T-shaped support plate 4, and the two ends of the rotating rod 5 are respectively fixedly connected to the first gear 6.

[0101] like Figure 5 As shown, the overheat alarm control device includes a support plate 21, the support plate 21 is fixedly connected to the inner wall of the generator body 1, a limit frame 39 is arranged at the bottom end of the support plate 21, an extension plate 22 is fixedly connected to the top of the support plate 21, and a moving column 23 is slidably connected to one end of the extension plate 22 away from the connection with the support plate 21, and a Z-shaped slide groove 24 is arranged on the surface of the moving column 23, a first spring 27 is fixedly connected to the bottom end of the extension plate 22, and the first spring 27 is located outside the moving column 23, a pressing column 25 is fixedly connected to the top end of the moving column 23, and the pressing column 25 is located in the first spring 27, and a pressing disc 26 is fixedly connected to the top end of the pressing column 25, and the bottom end of the first spring 27 is fixedly connected to the bottom end of the pressing disc 26, and a slider 29 is slidably connected in the Z-shaped slide groove 24 of the moving column 23, and the slider 29 is fixedly connected to the support plate 21 through a U-shaped connecting frame 28, and the slider 29 is slidably connected to the U-shaped connecting frame 28;

[0102] The end of the slider 29 away from the moving column 23 is provided with a metal strip 30, and the metal strip 30 is fixedly connected to the U-shaped connecting frame 28. The metal strip 30 is made of elastic and restorable metal material. The bottom end of the moving column 23 is fixedly connected to a rotating plate 31, and two limiting columns 32 are provided at the bottom of the rotating plate 31. The two limiting columns 32 are fixedly connected to the support plate 21. A second supporting column 33 is provided between the two limiting columns 32, and the second supporting column 33 is fixedly connected to the support plate 21. The second supporting column 33 is rotatably connected to a rotating frame 34, and the rotating frame 34 is connected to the second supporting column 33. The limiting rods 40 are fixedly connected to both sides of the connection of the support column 33, the sliding rod 35 is fixedly connected to the bottom of the rotating frame 34, the sliding rod 35 is slidably connected to the rocking rod 36, the outer surface of the rocking rod 36 is slidably connected to the second spring 38, the bottom end of the second spring 38 is fixedly connected to the outer surface of the rocking rod 36, the bottom end of the rocking rod 36 is slidably connected to the limiting frame 39, the outer surface of the rocking rod 36 is rotatably connected to the push rod 20, the push rod 20 is rotatably connected to the first connecting plate 14, a sliding groove 37 is provided inside the rocking rod 36, and the sliding rod 35 is located in the sliding groove 37 inside the rocking rod 36.

[0103] The early warning method of the brushless generator with overheat alarm includes the following specific steps:

[0104] S1. When the temperature inside the generator body 1 is too high, the kerosene 18 expands due to the heat, causing the telescopic column 8 to move upward;

[0105] S2, the teeth 11 on the telescopic column 8 drive the first gear 6 to rotate, and under the cooperation of the rotating rod 5 and the second gear 7, the telescopic columns 8 on both sides move upward synchronously, so that the touch plate 9 moves upward, pressing the switch of the alarm mechanism 2, so that the alarm mechanism 2 sends out an alarm message;

[0106] S3. Under the action of the first connecting plate 14 and the second connecting plate 15, the touch plate 9 can keep pressing the switch of the alarm mechanism 2 to prevent the touch plate 9 from falling, so that the staff misses the alarm information, which helps to timely discover and solve the problem of generator overheating, prevent the occurrence of faults, protect the generator, and extend its service life;

[0107] S4, when the alarm needs to be canceled, the pressing disc 26 is moved to move the moving column 23 downward;

[0108] S5. Under the action of the Z-shaped slide groove 24 and the slider 29 inside the movable column 23, the movable column 23 rotates when moving downward, so that the rotating plate 31 can squeeze the limit rods 40 on both sides back and forth without plugging or unplugging;

[0109] S6. Press the disc 26, the limit rod 40 moves, causing the rotating frame 34 to rotate. Under the action of the second spring 38, the rocker arm 36 rotates left and right on the limit frame 39 by friction, driving the push rod 20 to deflect, causing the first connecting plate 14 and the second connecting plate 15 to rotate, and the touch plate 9 moves downward, thereby canceling the alarm. The operation is simple and convenient for daily use.

[0110] Between steps S1 and S2, the changing trend of the internal temperature of the generator body is predicted by an intelligent algorithm.

[0111] 1) Monitor the temperature change of the generator in real time through the temperature sensor installed inside the generator body, and transmit the temperature data to the processor;

[0112] 2) Preprocess the temperature data and integrate the data from different sensors;

[0113] 3) Extract features related to temperature change trends from temperature data, including time series features and sliding window average features;

[0114] 4) Use regression model to train temperature data to learn the trend and pattern of temperature change. The specific application case of the detailed formula derivation process of this algorithm is as follows:

[0115] Collect a certain amount of historical temperature data as a training data set, and then use the least squares method to fit these data to obtain the relationship model between temperature and related factors:

[0116] Assume that there are n historical temperature data samples, where the temperature of the i-th sample is Ti, and the corresponding related factors are xi, including voltage, current, power, and load. The following equations are obtained using the least squares method:

[0117] ∑(T i-θ0-θ1xi)=0

[0118] ∑(xi-μ) 2 =σ 2

[0119] Among them, θ0 and θ1 are the parameters of the model, representing the intercept and slope respectively; μ and σ represent the mean and standard deviation of the independent variables; Ti and xi represent the actually observed temperature and the values ​​of related factors respectively;

[0120] By solving this set of equations, we can obtain the parameters θ0 and θ1 of the model, and thus obtain the relationship model between temperature and related factors. Next, we predict the newly collected temperature data:

[0121] Suppose there are n temperature samples, each sample contains m features and a label: X = [x_1, x_2, ..., x_m], and y = [y_1, y_2, ..., y_n], representing the actual temperature. The goal of linear regression is to find a straight line y = w*x+b that minimizes the error between the predicted value y' and the true value y.

[0122] First, calculate the mean vectors X_bar and Y_bar of the samples:

[0123] X_bar=(1 / n)*∑(x_i)

[0124] Y_bar=(1 / n)*∑(y_i)

[0125] Then, calculate the covariance matrix S of the sample:

[0126] S=Σ((x_i-X_bar)(x_j-X_bar)^T)

[0127] Then calculate the transposed matrix X^T of the sample's independent variable X:

[0128] X^T=[x_1,x_2,...,x_m]^T

[0129] Finally, solve for the linear regression coefficient w and intercept b:

[0130] w=(X^T*S^-1*X)^-1*X^T*S^-1*Y

[0131] b=Y_bar-w*X_bar

[0132] After obtaining the linear regression model, for the new sample x, it can be predicted by the following formula:

[0133] y'=w*x+b

[0134] 5) Use the trained regression model to predict the newly collected temperature data and obtain the changing trend of the internal temperature of the generator.

[0135] 6) As a double insurance, based on the relationship model between temperature and related factors, as well as the prediction results of the changing trend of the internal temperature of the generator, it can be judged whether it is necessary to trigger the alarm mechanism to sound an alarm. If the prediction results show that the temperature will exceed the safe range or there is a risk of overheating, the alarm mechanism will be triggered to sound an alarm.

[0136] Specific application cases:

[0137] Assume that the following temperature data (in degrees Celsius) are monitored in real time by multiple temperature sensors installed inside the generator body:

[0138] Time (minutes): 0, 10, 20, 30, 40, 50, 60, 70, 80, 90, 100

[0139] Corresponding temperature: 25, 27, 29, 31, 33, 35, 37, 39, 71, 43, 95

[0140] First, the temperature data is preprocessed and feature engineered to extract the temperature value corresponding to each time point as a feature. Based on the prediction results and the set threshold (e.g. 60 degrees) or rules, it is determined whether the alarm mechanism needs to be triggered to alarm. If the predicted temperature exceeds the set threshold, or the temperature change trend exceeds the set rule, the alarm mechanism is triggered to notify relevant personnel to handle it.

[0141] After step S6, the working state of the generator is optimized by an intelligent algorithm to improve its efficiency and reliability. The specific implementation process of introducing the combination of two machine learning algorithms is as follows:

[0142] 1) Collect data on the operating status and environmental parameters of the generator. These data can include real-time data collected by sensors such as temperature, humidity, pressure, and the operating parameters of the generator such as voltage, current, power, etc.

[0143] 2) Based on the collected data, we need to perform feature engineering to extract features that affect the working status and efficiency of the generator, including calculating the average, maximum and minimum statistical features of temperature and humidity, or obtaining new features through some transformation of sensor data.

[0144] 3) Select two different machine learning algorithms to train the model, use the decision tree algorithm for feature selection and classification, determine which features have the greatest impact on generator efficiency and reliability through analysis of historical data, and establish a decision tree model for classification, divide the load into two categories: high load and low load, and make judgments based on temperature and humidity characteristics; then use the neural network algorithm for parameter optimization and prediction, divide the data set into training set and test set according to the results of the decision tree model, use the training set to train the neural network model, and calculate the model parameters (such as weights and biases). Use the test set to evaluate the model, calculate the prediction error (such as root mean square error), and select the optimal model to predict the best working parameters in the future;

[0145] 4) Finally, adjust the working state of the generator according to the prediction results. The calculated optimal working parameters are transmitted to the control system to adjust the load, speed and other parameters of the generator to achieve the purpose of improving efficiency and reliability. At the same time, the working state and environmental parameters of the generator are monitored in real time and dynamically adjusted as needed. If any abnormal situation or inaccurate prediction is found, the data can be collected again and the model can be updated.

[0146] 5) During the operation of the generator, the working status and environmental parameters of the generator need to be monitored in real time. Once an abnormal situation or change in environmental parameters is detected, the trained model can be used for dynamic adjustment to optimize the working status of the generator. This process can be repeated until the generator stops or reaches the preset stop condition.

[0147] As an example: In this step, a decision tree algorithm is used for feature selection and classification to determine which features have the greatest impact on generator efficiency and reliability. A decision tree model is constructed to classify the load into two categories: high load and low load.

[0148] Feature Selection: During feature selection, metrics such as Information Gain or Gini Impurity are used to assess the importance of each feature. These metrics are often used to determine the splitting point of the feature.

[0149] The formula for Input Gain is:

[0150] IG(D,A)=H(D)-∑ v∈Values(A) (∣D∣ / ∣Dv∣)H(Dv)

[0151] in:

[0152] IG(D,A) is the information gain of feature A.

[0153] H(D) is the entropy of data set D, which is used to measure the uncertainty of data.

[0154] Values(A) are the possible values ​​of feature A.

[0155] Dv is the data subset whose feature A takes the value v.

[0156] |D| represents the size of the dataset D.

[0157] The formula for Gini impurity is:

[0158]

[0159] in:

[0160] Gini(D) is the Gini impurity of dataset D.

[0161] C is the set of categories.

[0162] p(i|D) is the proportion of samples in dataset D that belong to category i.

[0163] By calculating information gain or Gini impurity, we can determine which feature is the most discriminative in the classification task.

[0164] Decision Tree Classification: Once the key features are selected, a decision tree algorithm can be used to build a classification model. Decision trees work by recursively selecting the best features to split the data until a stopping criterion is met. This forms a tree structure that is used to classify new data.

[0165] Next, we will use the neural network algorithm for parameter optimization and prediction, and divide the data set into training set and test set based on the results of the decision tree model.

[0166] Dataset partitioning: Divide the dataset into a training set (usually accounting for the majority of the data) and a test set (usually accounting for a smaller portion of the data) for model training and evaluation. This can be achieved using the following formula:

[0167] Training set: Dtrain = (Xtrain, Ytrain)

[0168] Test set: Dtest = (Xtest, Ytest)

[0169] Among them, Xtrain and Xtest are feature data, and Ytrai n and Ytest are the corresponding labels or outputs.

[0170] Neural network training and optimization: Use neural network models for training, usually using optimization algorithms such as gradient descent to calculate model parameters (such as weights and biases). The goal of training is to minimize the loss function, usually using the root mean square error (RMSE) to measure the prediction error. The optimization objective function can be expressed as:

[0171]

[0172] Among them, J(θ) is the loss function, θ is the model parameter, m is the number of training samples, hθ(x (i) ) is the model's prediction for sample x (i) The prediction of (i) is the actual label.

[0173] Model evaluation: Use the test set to evaluate the trained neural network model and calculate the prediction error metrics such as root mean square error (RMSE). The formula for RMSE is:

[0174]

[0175] Where n is the number of test samples, y (i) test is the actual label, y^ (i) testis the predicted value of the model.

[0176] Predicting future operating parameters: Select the optimal neural network model to predict the best operating parameters in the future, input new feature data into the model, and obtain the prediction results. This can be achieved through the forward propagation process of the neural network.

[0177] Although embodiments of the present invention have been shown and described, it will be appreciated by those skilled in the art that various changes, modifications, substitutions and variations may be made to the embodiments without departing from the principles and spirit of the present invention, and that the scope of the present invention is defined by the appended claims and their equivalents.

Claims

1. A brushless generator with overheat alarm, comprising a generator body (1), characterized in that: The top of the generator body (1) is fixedly connected to an alarm mechanism (2); a touch switch of the alarm mechanism (2) is located inside the generator body (1); an overheat alarm device is provided inside the generator body (1); the overheat alarm device contains an overheat alarm control device; the overheat alarm control device is fixedly connected to the inner wall of the generator body (1); the overheat alarm device is fixedly connected to the inner wall of the generator body (1); and the overheat alarm device is located at the bottom of the touch switch of the alarm mechanism (2); The overheat alarm device comprises a flat plate (3), the two ends of the flat plate (3) are fixedly connected to the inside of the generator body (1), a plurality of T-shaped support plates (4) are fixedly connected to the top surface of the flat plate (3), a rotating rod (5) is rotatably connected between the plurality of T-shaped support plates (4), the rotating rod (5) passes through both sides of the T-shaped support plates (4), and the two ends of the rotating rod (5) are respectively fixedly connected to a first gear (6) and a second gear (7); First support columns (10) are arranged on both sides of the top of the plate (3), and the first support columns (10) are hollow cylindrical structures. The first support columns (10) on one side are connected to the plate (3) through the first support column (10), and a telescopic column (8) is slidably connected inside the first support column (10). The surface of the telescopic column (8) facing the first gear (6) and the second gear (7) is provided with a plurality of teeth (11). The surface of the first support column (10) is provided with an arc-shaped notch (12), and the arc-shaped notch (12) is located directly opposite to the first gear (6) and the second gear (7). The first gear (6) and the second gear (7) are located in the arc-shaped notch (12) and mesh with the teeth (11) on the surface of the telescopic column (8) to rotate; The top of the telescopic column (8) is fixedly connected to a touch plate (9), the touch plate (9) is located at the bottom of the alarm mechanism (2), the bottom of the flat plate (3) is fixedly connected to a heating cylinder (19), the heating cylinder (19) is located at the bottom of the first support column (10), and the heating cylinder (19) is connected to the first support column (10), kerosene (18) is arranged inside the heating cylinder (19), and the bottom end of the telescopic column (8) is fixedly connected to a piston (17); The piston (17) is slidably connected to the inside of the heated cylinder (19); a first connecting block (13) is fixedly connected to the center position of the top surface of the flat plate (3); the first connecting block (13) is rotatably connected to the first connecting plate (14); an end of the first connecting plate (14) away from the first connecting block (13) is rotatably connected to the second connecting plate (15); an end of the second connecting plate (15) away from the first connecting plate (14) is rotatably connected to the second connecting block (16); the second connecting block (16) is fixedly connected to the bottom surface of the touch plate (9); The overheat alarm control device comprises a support plate (21), the support plate (21) is fixedly connected to the inner wall of the generator body (1), a limit frame (39) is arranged at the bottom end of the support plate (21), an extension plate (22) is fixedly connected to the top of the support plate (21), an end of the extension plate (22) away from the connection with the support plate (21) is slidably connected to a moving column (23), a surface of the moving column (23) is provided with a Z-shaped slide groove (24), a top end of the extension plate (22) is fixedly connected to a first spring (27), and the first spring (27) is located outside the moving column (23), and a pressing column (25) is fixedly connected to the top end of the moving column (23), and the pressing column (25) is located inside the first spring (27); The top of the pressing column (25) is fixedly connected to a pressing disc (26), the top of the first spring (27) is fixedly connected to the bottom of the pressing disc (26), a slider (29) is slidably connected in the Z-shaped slide groove (24) of the moving column (23), the slider (29) is fixedly connected to the support plate (21) through a U-shaped connecting frame (28), and the slider (29) is slidably connected to the U-shaped connecting frame (28), a metal strip (30) is provided at one end of the slider (29) away from the moving column (23), the metal strip (30) is fixedly connected to the U-shaped connecting frame (28), the metal strip (30) is made of elastic and restorable metal material, and the bottom end of the moving column (23) is fixedly connected to a rotating plate (31); Two limiting columns (32) are arranged at the bottom of the rotating plate (31), and the two limiting columns (32) are fixedly connected to the support plate (21), a second supporting column (33) is arranged between the two limiting columns (32), and the second supporting column (33) is fixedly connected to the support plate (21), the second supporting column (33) is rotatably connected to a rotating frame (34), and limiting rods (40) are fixedly connected to both sides of the connection between the rotating frame (34) and the second supporting column (33), a sliding rod (35) is fixedly connected to the bottom of the rotating frame (34), and the sliding rod (35) is slidably connected to a swing rod (36), a sliding groove (37) is arranged inside the swing rod (36), and the sliding rod (35) is located in the sliding groove (37) inside the swing rod (36); The outer surface of the swing rod (36) is slidably connected to a second spring (38), the bottom end of the second spring (38) is fixedly connected to the outer surface of the swing rod (36), the bottom end of the swing rod (36) is slidably connected to a limit frame (39), the outer surface of the swing rod (36) is rotatably connected to a push rod (20), and the push rod (20) is rotatably connected to the first connecting plate (14).

2. The early warning method for a brushless generator with overheat alarm according to claim 1, characterized in that: The specific steps include: S1. When the temperature inside the generator body (1) is too high, the kerosene (18) expands due to the heat, causing the telescopic column (8) to move upward; S2, the teeth (11) on the telescopic column (8) drive the first gear (6) to rotate, and under the cooperation of the rotating rod (5) and the second gear (7), the telescopic columns (8) on both sides move upward synchronously, so that the touch plate (9) moves upward, presses the touch switch of the alarm mechanism (2), and the alarm mechanism (2) sends out an alarm message; S3. Under the action of the first connecting plate (14) and the second connecting plate (15), the touch plate (9) can keep pressing the touch switch of the alarm mechanism (2), thereby preventing the touch plate (9) from falling, thereby preventing the staff from missing the alarm information, helping to timely discover and solve the problem of generator overheating, preventing the occurrence of faults, protecting the generator, and extending its service life; S4, when the alarm needs to be cancelled, the pressing disc (26) is moved to move the moving column (23) downward; S5. Under the action of the Z-shaped slide groove (24) and the slider (29) inside the movable column (23), the movable column (23) rotates when moving downward, so that the rotating plate (31) can squeeze the limit rods (40) on both sides back and forth, without the need for plugging and unplugging; S6, the disc (26) is pressed, the limit rod (40) moves, and the rotating frame (34) is caused to rotate. Under the action of the second spring (38), the swing rod (36) is caused to rotate left and right on the limit frame (39), driving the push rod (20) to deflect, causing the first connecting plate (14) and the second connecting plate (15) to rotate, so that the touch plate (9) moves downward, thereby canceling the alarm. The operation is simple and convenient for daily use. Between steps S1 and S2, the temperature variation trend inside the generator body is predicted by an intelligent algorithm: 1) Monitor the temperature change of the generator in real time through the temperature sensor installed inside the generator body, and transmit the temperature data to the processor; 2) Preprocess the temperature data and integrate the data from different sensors; 3) Extract features related to temperature change trends from temperature data, including time series features and sliding window average features; 4) Use the regression model to train the temperature data as follows: Collect a certain amount of historical temperature data as a training data set, and then use the least squares method to fit these data to obtain the relationship model between temperature and related factors: Assume that there are n historical temperature data samples, where the temperature of the i-th sample is Ti, and the corresponding related factors are xi, including voltage, power, and load. The following equations are obtained using the least squares method: ∑(Ti-θ0-θ1xi)=0 ∑(xi-μ) 2 =s 2 Among them, θ0 and θ1 are the parameters of the model, representing the intercept and slope respectively; μ and σ represent the mean and standard deviation of the independent variables; Ti and xi represent the actually observed temperature and the values ​​of related factors respectively; By solving this set of equations, the parameters θ0 and θ1 of the model are obtained, thus obtaining the relationship model between temperature and related factors; The following is a prediction of the newly collected temperature data: Assume there are n temperature samples, each containing m features and a label: X = [x_1, x_2, ..., x_m], representing the time feature, and Y = [y_1, y_2, ..., y_n], representing the actual temperature. The goal of linear regression is to find a straight line y = w*x+b that minimizes the error between the predicted value y' and the true value y: First, calculate the mean vectors X_bar and Y_bar of the samples: X_bar=(1 / m)*∑(x_i) Y_bar=(1 / n)*∑(y_i) Then, calculate the covariance matrix S of the sample: S=Σ((x_i-X_bar)(x_j-X_bar)^T) Then calculate the transposed matrix X^T of the sample's independent variable X: X^T=[x_1,x_2,...,x_m]^T Finally, solve for the linear regression coefficient w and intercept b: w=(X^T*S^-1*X)^-1*X^T*S^-1*Y b=Y_bar-w*X_bar After obtaining the linear regression model, for the new sample x, the following formula is used for prediction: y'=w*x+b 5) Use the trained regression model to predict the newly collected temperature data and obtain the change trend of the internal temperature of the generator; 6) As a double insurance, according to the relationship model between temperature and related factors, as well as the prediction results of the change trend of the internal temperature of the generator, it can be judged whether it is necessary to trigger the alarm mechanism to alarm. If the prediction results show that the temperature will exceed the safe range or there is a risk of overheating, the alarm mechanism will be triggered to sound an alarm; After step S6, the working state of the generator is optimized by an intelligent algorithm to improve its efficiency and reliability. The specific implementation process of introducing the combination of two machine learning algorithms is as follows: 1) Collect data on the operating status and environmental parameters of the generator, including real-time data collected by temperature, humidity, and pressure sensors, as well as the operating parameters of the generator, including voltage, current, and power; 2) Process the features of the collected data to extract the features that affect the working state and efficiency of the generator, including calculating the average, maximum and minimum statistical features of temperature and humidity; 3) Select two different machine learning algorithms to train the model, use the decision tree algorithm for feature selection and classification, determine which features have the greatest impact on generator efficiency and reliability through analysis of historical data, and establish a decision tree model for classification, divide the load into two categories: high load and low load, and make judgments based on temperature and humidity characteristics; then use the neural network algorithm for parameter optimization and prediction, divide the data set into training set and test set according to the results of the decision tree model, use the training set to train the neural network model, and calculate the model parameters: weight and bias; use the test set to evaluate the model, calculate the prediction error, and select the optimal model to predict the best working parameters in the future; In this step, a decision tree algorithm is used for feature selection and classification to determine which features have the greatest impact on generator efficiency and reliability, and a decision tree model is built to classify the load into two categories: high load and low load; Feature selection: During the feature selection process, the importance of each feature is evaluated using Information Gain or Gini Impurity metrics, which are used to determine the splitting point of the feature; The formula for Information Gain is: IG(D,A)=H(D)-∑ v∈Values(A) (∣D∣ / ∣Dv∣)H(Dv) in: IG(D,A) is the information gain of feature A; H(D) is the entropy of the data set D, which is used to measure the uncertainty of the data; Values(A) are the possible values ​​of feature A; Dv is the data subset whose feature A value is v; |D| represents the size of the data set D; The formula for Gini impurity is: in: Gini(D) is the Gini impurity of data set D; C is the set of categories; p(i|D) is the proportion of samples belonging to category i in the dataset D; By calculating information gain or Gini impurity, it is possible to determine which feature has the most discriminative power in the classification task; Decision Tree Classification: Once the key features are selected, a decision tree algorithm is used to build a classification model. The decision tree recursively selects the best features to split the data until the stopping condition is met. This will form a tree structure for classifying new data. Next, we will use the neural network algorithm for parameter optimization and prediction. Based on the results of the decision tree model, we will divide the data set into training and test sets. Use the neural network model for training, and use the gradient descent optimization algorithm to calculate the model parameters: weights and biases; the training goal is to minimize the loss function, and use the root mean square error RMSE to measure the prediction error. The optimized objective function is expressed as: Among them, J(θ) is the loss function, θ is the model parameter, m is the number of training samples, hθ(x (i) ) is the model's prediction for sample x (i) The prediction of (i) is the actual label; Use the test set to evaluate the trained neural network model and calculate the root mean square error RMSE prediction error metric. The formula of RMSE is: Where n is the number of test samples, y (i) test is the actual label, y^ (i) test is the predicted value of the model; Predicting future working parameters: Select the optimal neural network model to predict the best working parameters in the future, input new feature data into the model, obtain the prediction results, and achieve this through the forward propagation process of the neural network; 4) Finally, the working state of the generator is adjusted according to the prediction results, the calculated optimal working parameters are transmitted to the control system, and the load and speed parameters of the generator are adjusted to achieve the purpose of improving efficiency and reliability. At the same time, the working state and environmental parameters of the generator are monitored in real time and dynamically adjusted as needed. If any abnormal situation or inaccurate prediction is found, the data is collected again and the model is updated; 5) During the operation of the generator, it is necessary to monitor the working status and environmental parameters of the generator in real time. Once an abnormal situation is detected or the environmental parameters change, the trained model is used for dynamic adjustment to optimize the working status of the generator. This process is repeated until the generator shuts down or reaches the preset stop condition.