An energy management method based on comprehensive performance evaluation and power demand forecasting

Abstract

The present invention relates to an energy management method based on comprehensive performance evaluation and power demand prediction, including: S1, establishing a comprehensive evaluation index system for hybrid welding robots, and calculating the evaluation level; S2, using the evaluation level as a robot operation state level factor, obtaining lithium battery state information, judging the system's working mode according to the relationship between the fuel cell power and the robot power at the next moment, and obtaining a battery power allocation plan; S3, obtaining the proton exchange membrane fuel cell power according to the state information and output strategy of the input variables of the energy management controller; S4, establishing a power demand prediction model through a random forest method; S5, obtaining the load power demand of the welding robot according to the model, and controlling the robot to complete welding according to the battery power allocation plan. Compared with the prior art, the present invention has the advantages of improving the energy utilization rate of the robot system, ensuring the stability of the power supply, and improving the stability of the robot's operating speed.

Description

Technical Field

[0001] The present invention relates to the technical field of robot power transmission, and in particular to an energy management method based on comprehensive performance evaluation and power demand prediction. Background Art

[0002] With the development of science and technology, large ships and cross-sea bridges are constantly being built. They not only carry large capacity, but also undertake important tasks. These large structural parts work in high humidity and corrosive environments all year round, which is a very important task and bears huge loads. Therefore, mobile robots with high-quality automatic welding or detection are needed. The precision and high accuracy of the robot require the driving energy to have the characteristics of fast starting and braking, stable operation, and long continuous working time. At present, cable power supply limits the continuous automatic operation of such robots over a large range and long distance. Although driving energy such as lithium batteries can realize the free movement of robots, they still need to be charged. Not only is the charging time long, but also the power consumption economy is limited due to the influence of charging efficiency. In addition, the impact of aging on the performance and economy of lithium batteries is an obstacle to the expansion of lithium battery applications.

[0003] In order to better solve the above problems existing in the power system of the welding robot and enable the welding robot to perform various tasks more quickly, effectively and stably, it is necessary to study an energy management strategy that meets the requirements of high precision, real-time performance, stability and energy economy. Summary of the invention

[0004] The purpose of the present invention is to overcome the defects of the above-mentioned prior art and to provide an energy management method based on comprehensive performance evaluation and power demand prediction, so that the welding robot can perform tasks more quickly, effectively and stably, and meet the requirements of high precision, real-time, stability and energy economy.

[0005] The purpose of the present invention can be achieved by the following technical solutions:

[0006] An energy management method based on comprehensive performance evaluation and power demand prediction, which comprehensively evaluates the energy control system through a cloud model and applies the comprehensive evaluation to the energy management system, specifically includes the following steps:

[0007] S1. Establish a comprehensive evaluation index system for hybrid welding robots and calculate the evaluation level;

[0008] S2. Using the evaluation level as the robot operation status level factor, obtaining the lithium battery status information, judging the working mode of the hybrid power system according to the relationship between the fuel cell power and the predicted robot power at the next moment, and calculating the distribution plan of the fuel cell power and the lithium battery power at the next moment in combination with the operation status evaluation level factor;

[0009] S3, according to the acquired state information of the input variables of the energy management controller, the power of the proton exchange membrane fuel cell is calculated according to the preset output strategy;

[0010] S4. According to the power of proton exchange membrane fuel cell, a power demand prediction model is established by using the random forest method;

[0011] S5. Obtain the load power demand of the welding robot under corresponding working conditions according to the power demand prediction model, control the welding robot to complete welding according to the distribution scheme of fuel cell power and lithium battery power, and track the welding trajectory of the welding robot.

[0012] The comprehensive evaluation index system of the hybrid power welding robot is provided with a reliability index, a stability index and an economic index.

[0013] Furthermore, the energy control system of the fuel cell hybrid mobile robot takes the completion of the designed goal as the primary task, and the reliability indicators include seam tracking error, charge state and unidirectional DC / DC output voltage.

[0014] Furthermore, while ensuring that the system completes its tasks, it is also necessary to achieve smooth, orderly and safe operation, and the stability indicators include PEMFC output power, lithium battery current, left wheel acceleration, right wheel acceleration and welding torch acceleration.

[0015] Furthermore, considering the main energy consumption and energy utilization of the system, the economic indicators include fuel cell efficiency, H2 utilization and load power ratio.

[0016] The hybrid welding robot comprehensive evaluation index system comprehensively evaluates the operating status of the entire energy control system according to the entropy weight method and the cloud model to obtain a rating grade.

[0017] The output strategy in step S3 includes a low state of charge interval, a normal state of charge interval and a high state of charge interval.

[0018] Furthermore, the low state of charge interval, normal state of charge interval and high state of charge interval all include a robot power state where the fuel cell power is less than the robot power state at the next moment, a robot power state where the fuel cell power is equal to the robot power state at the next moment and a robot power state where the fuel cell power is greater than the robot power state at the next moment.

[0019] Furthermore, in the low state of charge interval:

[0020] If the fuel cell power is greater than the robot power state at the next moment, the fuel cell provides the lowest power in the optimal output range, and the lithium battery is charged faster;

[0021] If the fuel cell power is equal to the robot power state at the next moment, the battery power is low at this time, the PEMFC output power increases, and the lithium battery is charged faster;

[0022] If the fuel cell power is less than the robot power state at the next moment, the battery power is very low at this time. In order to protect the battery, it is necessary to increase the PEMFC output power as soon as possible to relieve the power pressure of the lithium battery.

[0023] Furthermore, in the normal state of charge interval:

[0024] If the fuel cell power is greater than the robot power state at the next moment, the PEMFC power is equal to the power P change1 Variation amplitude, P change1 The robot's changing speed is dynamically adjusted according to its status evaluation level;

[0025] If the fuel cell power is equal to the robot power state at the next moment, keep the current power of the fuel cell unchanged;

[0026] If the fuel cell power is less than the robot power state at the next moment, the PEMFC power is equal to the power P change1 Variation amplitude, P change1 The speed of change is dynamically adjusted by the robot's status evaluation level.

[0027] Furthermore, in the high state of charge interval:

[0028] If the fuel cell power is greater than the robot power state at the next moment, the PEMFC output power is accelerated to decrease until it is less than the load power, providing output conditions for the lithium battery;

[0029] If the fuel cell power is equal to the robot power state at the next moment, reduce the output power of the PEMFC and increase the output power ratio of the lithium battery;

[0030] If the fuel cell power is less than the robot power state at the next moment, the PEMFC output power is reduced and the output power of the lithium battery continues to increase.

[0031] In step S4, the random forest method randomly selects multiple groups of training sample subsets from the input working condition training samples through the autonomous sampling method to construct respective CART decision trees, and trains the regression tree through the random subspace method to obtain the power demand prediction results of each decision tree, and obtains the final prediction result by using the simple averaging method to calculate the average value of the prediction results of all decision trees.

[0032] The parameters of the power demand prediction model are of a mixed type including integer values ​​and floating point values, and a hybrid particle swarm optimization algorithm is used to train the parameters of the model.

[0033] Compared with the prior art, the present invention has the following beneficial effects:

[0034] 1. The present invention applies comprehensive evaluation to the energy management system. In order to achieve the goal of optimizing the overall stability and economy of the robot, it is determined to focus on optimizing the power supply performance of the robot, better meet the power demand of the mobile welding robot, realize the reasonable distribution of fuel cell battery power, give full play to the efficiency of the hybrid power system, and enable the mobile welding robot to operate smoothly and efficiently under various working conditions to obtain better performance.

[0035] 2. The mobile welding robot driving system and driving method of the present invention adopt a comprehensive evaluation method to evaluate the welding robot and perform reasonable power allocation, have better dynamic response performance to drastically changing load conditions, give full play to the effectiveness of the hybrid power system, and enable the welding robot to operate smoothly and efficiently under various working conditions to obtain better performance.

[0036] 3. The present invention adopts random forest to predict the load power of the welding robot, sets the model parameters with the minimum mean square error through hybrid particle swarm optimization, and obtains a high-precision random forest prediction model.

[0037] 4. The hybrid power system energy management rules established by the present invention based on comprehensive evaluation improve the stability and economy of the welding robot, thereby further improving the welding quality of the welding robot. BRIEF DESCRIPTION OF THE DRAWINGS

[0038] Figure 1 It is a schematic diagram of the process of the present invention;

[0039] Figure 2 A system block diagram of robot system performance optimization based on evaluation level and power demand prediction according to the present invention;

[0040] Figure 3 A schematic diagram of the energy output strategy of the fuel cell hybrid welding robot of the present invention;

[0041] Figure 4 It is a curve diagram of the welding seam tracking trajectory of the welding robot of the present invention;

[0042] Figure 5 A schematic diagram of load power distribution of the robot of the present invention;

[0043] Figure 6 Schematic diagram of the charge state change curve of the lithium battery of the present invention;

[0044] Figure 7 This is a comparison chart of the output power of the fuel cell of the present invention. DETAILED DESCRIPTION

[0045] The present invention is described in detail below in conjunction with the accompanying drawings and specific embodiments. This embodiment is implemented based on the technical solution of the present invention, and provides a detailed implementation method and specific operation process, but the protection scope of the present invention is not limited to the following embodiments.

[0046] Example

[0047] like Figure 1 As shown, an energy management method based on comprehensive performance evaluation and power demand prediction is used to comprehensively evaluate the energy control system through a cloud model, and the comprehensive evaluation is applied to the energy management system. The energy control system includes a PEMFC fuel cell, a lithium battery, a unidirectional boost (Boost) DC / DC converter, a bidirectional buck-boost DC / DC converter, and an energy management system (EMS). The PEMFC serves as the main power source, and its output end is connected to the DC bus through a Boost DC / DC converter. The lithium battery serves as an auxiliary power source and is connected to the DC bus in parallel with the PEMFC through a bidirectional DC / DC converter. The energy management system is responsible for detecting signals such as the voltage and current of each unit, and outputs the control signal of the DC / DC converter according to the set research management strategy, so as to distribute power. The method specifically includes the following steps:

[0048] S1. Establish a comprehensive evaluation index system for hybrid welding robots and calculate the evaluation level;

[0049] S2. Using the evaluation level as the robot operation status level factor, obtaining the lithium battery status information, judging the working mode of the hybrid power system according to the relationship between the fuel cell power and the predicted robot power at the next moment, and calculating the distribution plan of the fuel cell power and the lithium battery power at the next moment in combination with the operation status evaluation level factor;

[0050] S3, according to the acquired state information of the input variables of the energy management controller, the power of the proton exchange membrane fuel cell is calculated according to the preset output strategy;

[0051] S4. According to the power of proton exchange membrane fuel cell, a power demand prediction model is established by using the random forest method;

[0052] S5. Obtain the load power demand of the welding robot under corresponding working conditions according to the power demand prediction model, control the welding robot to complete welding according to the distribution scheme of fuel cell power and lithium battery power, and track the welding trajectory of the welding robot.

[0053] The comprehensive evaluation index system of hybrid welding robots includes reliability index, stability index and economic index.

[0054] The energy control system of the fuel cell hybrid mobile robot takes the completion of the design objectives as its primary task, and the reliability indicators include seam tracking error, charge state and unidirectional DC / DC output voltage.

[0055] While ensuring that the system completes its mission, it must also achieve smooth, orderly and safe operation. Stability indicators include PEMFC output power, lithium battery current, left wheel acceleration, right wheel acceleration and welding torch acceleration.

[0056] Considering the main energy consumption and energy utilization of the system, the economic indicators include fuel cell efficiency, H2 utilization and load power ratio.

[0057] The comprehensive evaluation index system of hybrid welding robots comprehensively evaluates the operating status of the entire energy control system based on the entropy weight method and cloud model to obtain a rating grade.

[0058] like Figure 3 As shown, the output strategy in step S3 includes a low state of charge interval, a normal state of charge interval and a high state of charge interval.

[0059] The low state of charge interval, normal state of charge interval and high state of charge interval all include the fuel cell power P fc (k) is less than the robot power state P at the next moment fc (k+1), the fuel cell power is equal to the robot power state at the next moment and the fuel cell power is greater than the robot power state at the next moment.

[0060] In the low state of charge range:

[0061] If the fuel cell power is greater than the robot power state at the next moment, the fuel cell provides the lowest power in the optimal output range, and the lithium battery is charged faster;

[0062] If the fuel cell power is equal to the robot power state at the next moment, the battery power is low at this time, the PEMFC output power increases, and the lithium battery is charged faster;

[0063] If the fuel cell power is less than the robot power state at the next moment, the battery power is very low at this time. In order to protect the battery, it is necessary to increase the PEMFC output power as soon as possible to relieve the power pressure of the lithium battery.

[0064] In the normal state of charge range:

[0065] If the fuel cell power is greater than the robot power state at the next moment, the PEMFC power is equal to the power P change1 Variation amplitude, P change1 The robot's changing speed is dynamically adjusted according to its status evaluation level;

[0066] If the fuel cell power is equal to the robot power state at the next moment, keep the current power of the fuel cell unchanged;

[0067] If the fuel cell power is less than the robot power state at the next moment, the PEMFC power is equal to the power P change1 Variation amplitude, P change1 The speed of change is dynamically adjusted by the robot's status evaluation level.

[0068] In the high state of charge range:

[0069] If the fuel cell power is greater than the robot power state at the next moment, the PEMFC output power is accelerated to decrease until it is less than the load power, providing output conditions for the lithium battery;

[0070] If the fuel cell power is equal to the robot power state at the next moment, reduce the output power of the PEMFC and increase the output power ratio of the lithium battery;

[0071] If the fuel cell power is less than the robot power state at the next moment, the PEMFC output power is reduced and the output power of the lithium battery continues to increase.

[0072] In step S4, the random forest method randomly selects multiple groups of training sample subsets from the input working condition training samples through the autonomous sampling method to construct respective CART decision trees, and trains the regression tree through the random subspace method to obtain the power demand prediction results of each decision tree, and obtains the final prediction result by using the simple averaging method to calculate the average value of the prediction results of all decision trees.

[0073] The parameters of the power demand prediction model are a mixed type including integer values ​​and floating-point values. The hybrid particle swarm optimization algorithm is used to train the parameters of the model.

[0074] In specific implementation, the mobile welding robot maintains a stable posture during the tracking of the weld. Figure 4 The weld seam tracking trajectory of the mobile welding robot with a broken line shape and an included angle of 120° is tracked.

[0075] like Figure 5As shown in the figure, the first 10 seconds is the startup phase of the robot. The output power of the fuel cell at startup cannot meet the high power demand of the load, and the load power is in a state of strong fluctuation. At this time, the strategy allocates the main load power to the lithium battery. The PEMFC output power is rapidly increasing and close to the load power, and the output power of the lithium battery is gradually decreasing. The difference between the load power and the PEMFC output power determines the speed of its power change, and the level of comprehensive evaluation of the robot performance has been corrected to allow the PEMFC output power to change significantly when the robot's operating status level is high (the system's comprehensive performance is high and the state is stable). The proposed robot performance optimization strategy sets a new P with the fluctuation of power demand under the principle of stabilizing the fuel cell output and reducing drastic fluctuations. fc (k+1)) value. Figure 5 The output power of PEMFC follows the load power demand to a certain extent, but when the load power fluctuates violently, its output power curve remains relatively stable, and the power shortage caused by the fluctuation is supplemented by the lithium battery. Therefore, during the load power fluctuation period, the output power curve of the lithium battery and the load power curve have similar complex fluctuations as a whole, and the waveform of the fuel cell output curve at this time is relatively stable. The change of the lithium battery state of charge (SOC) during this stage also reflects this phenomenon.

[0076] like Figure 6 As shown in the figure, the SOC curve of the lithium battery fluctuates frequently around 100s, which is caused by the fact that the load power changes quickly and the PEMFC output power follows with hysteresis. Therefore, there will be a situation where the fuel cell and the lithium battery are powered together, and the fuel cell is powered while the lithium battery is charged alternately. At this time, the SOC value of the lithium battery will constantly change in the trend of increase and decrease.

[0077] like Figure 7 As shown in the figure, the fuel cell output power curve based on power demand prediction and evaluation level strategy has better performance in the high-frequency fluctuation area compared with the power output curve based on actual power strategy, and its curve is smoother. And the standard deviation of 38.08 based on power demand prediction + evaluation level strategy is much smaller than 42.92 based on actual power strategy, which is reduced by nearly 11.26%. Obviously, the fuel cell output power fluctuation of the proposed method is smoother and has good output stability. In addition, stable fuel cell output can significantly delay its performance degradation, ensure its operating performance, extend its service life, reduce the use cost of robot hardware, and thus improve the economy of the robot.

[0078] From the experimental results of this embodiment, it can be seen that the method proposed by the present invention can better optimize a series of parameters of the robot system, reasonably distribute the output load of PEMFC and lithium battery under the condition of complex dynamic changes in load power, improve the energy utilization rate of the robot system, ensure the stability of power supply, and improve the stability of the robot's running speed, so that the tracking accuracy of the robot is also slightly improved. Therefore, the present invention can improve the stability and economy of the robot system and make it have better performance.

[0079] In addition, it should be noted that the specific embodiments described in this specification may be named differently, and the above contents described in this specification are merely examples of the structure of the present invention. Any equivalent changes or simple changes made based on the structure, features and principles of the present invention are included in the protection scope of the present invention. Those skilled in the art of the present invention may make various modifications or supplements to the specific examples described or adopt similar methods, as long as they do not deviate from the structure of the present invention or exceed the scope defined by the claims, they should all fall within the protection scope of the present invention.

Claims

1. An energy management method based on comprehensive performance evaluation and power demand prediction, characterized in that: The energy control system is comprehensively evaluated through the cloud model, which includes the following steps: S1. Establish a comprehensive evaluation index system for hybrid welding robots and calculate the evaluation level; S2. Using the evaluation level as the robot operation status level factor, obtaining the lithium battery status information, judging the working mode of the hybrid power system according to the relationship between the fuel cell power and the predicted robot power at the next moment, and calculating the distribution plan of the fuel cell power and the lithium battery power at the next moment in combination with the operation status evaluation level factor; S3, according to the acquired state information of the input variables of the energy management controller, the power of the proton exchange membrane fuel cell is calculated according to the preset output strategy; S4. According to the power of proton exchange membrane fuel cell, a power demand prediction model is established by using the random forest method; S5. Obtain the load power demand of the welding robot under corresponding working conditions according to the power demand prediction model, control the welding robot to complete welding according to the distribution scheme of the fuel cell power and the lithium battery power, and track the welding trajectory of the welding robot; The comprehensive evaluation index system of the hybrid welding robot includes reliability index, stability index and economic index; The hybrid welding robot comprehensive evaluation index system comprehensively evaluates the operating status of the entire energy control system according to the entropy weight method and the cloud model to obtain a rating grade.

2. The energy management method based on comprehensive performance evaluation and power demand prediction according to claim 1, characterized in that: The reliability indicators include seam tracking error, charge state and unidirectional DC / DC output voltage.

3. The energy management method based on comprehensive performance evaluation and power demand prediction according to claim 1, characterized in that: The stability indicators include PEMFC output power, lithium battery current, left wheel acceleration, right wheel acceleration and welding torch acceleration.

4. The energy management method based on comprehensive performance evaluation and power demand prediction according to claim 1, characterized in that: The economic indicators include fuel cell efficiency, H2 utilization rate and load power ratio.

5. The energy management method based on comprehensive performance evaluation and power demand prediction according to claim 1, characterized in that: The output strategy in step S3 includes a low state of charge interval, a normal state of charge interval and a high state of charge interval.

6. The energy management method based on comprehensive performance evaluation and power demand prediction according to claim 5, characterized in that: The low state of charge interval, normal state of charge interval and high state of charge interval all include a robot power state where the fuel cell power is less than the robot power state at the next moment, a robot power state where the fuel cell power is equal to the robot power state at the next moment and a robot power state where the fuel cell power is greater than the robot power state at the next moment.

7. The energy management method based on comprehensive performance evaluation and power demand prediction according to claim 1, characterized in that: In step S4, the random forest method randomly selects multiple groups of training sample subsets from the input working condition training samples through the autonomous sampling method to construct respective CART decision trees, and trains the regression tree through the random subspace method to obtain the power demand prediction results of each decision tree, and obtains the final prediction result by using the simple averaging method to calculate the average value of the prediction results of all decision trees.

8. The energy management method based on comprehensive performance evaluation and power demand prediction according to claim 1, characterized in that: The parameters of the power demand prediction model are of a mixed type including integer values ​​and floating point values, and a hybrid particle swarm optimization algorithm is used to train the parameters of the model.

Images