Pump speed control method and device, computer program and computer readable medium having stored thereon the computer program used thereby, and pump

Abstract

The present invention describes a method and apparatus for controlling the speed of a pump with a variable speed drive in a cyclic application, the cyclic application comprising repeated cycles, each cycle consisting of a pump-down phase and a holding phase, wherein the start time point of each cycle is the time point when the rise in the inlet pressure of the pump is sufficiently large, and thus the time extending between two consecutive cycle start time points is the cycle time. The control method comprises determining the start of the next cycle during the current cycle, wherein preferably the current cycle directly precedes the next cycle. The method further comprises, before the start of the next cycle, controlling the pump to accelerate to the maximum allowed speed during the holding phase of the current cycle, so that at the start of the next cycle, the full pump capacity is available.

Description

Technical Field

[0001] The present invention relates to a method for controlling the speed of a pump, in particular but not exclusively a variable speed drive (VSD) vacuum pump employed in circulation applications. Furthermore, the present invention relates to an apparatus for performing such a pump speed control method. Furthermore, the present invention relates to a pump coupled to such a pump speed control apparatus. Background Art

[0002] Pumping systems have been widely used in many sectors in industry, however, due to low energy efficiency, pumping systems consume a lot of energy. For example, a constant speed pump operates at a constant high pumping speed just to maintain a low capacity of the pump at the same time. Moreover, in actual application scenarios, it is rare that the requirements of the pump user will remain unchanged during the pumping process. Therefore, pumps driven by VSDs come into play, because it is possible to slow down the pump while maintaining satisfactory performance of the pumping system, resulting in energy savings.

[0003] For example, a VSD vacuum pump is used in an application that requires repeated cyclic operation of a pumping system. Such a cyclic application is usually composed of two stages, namely a pump down stage and a holding stage. In the pump down stage, the pump is kept operating at maximum speed so that the pressure in the vacuum chamber is reduced to the desired target pressure, after which the pump speed is reduced to maintain the desired target pressure. Therefore, the energy consumption in the pump down stage is the largest, and the time required to pump down to the desired target pressure is already the best time that the pump can deliver. In the holding stage, the VSD pump is kept at the maximum speed, resulting in a waste of energy but the best pump down for the next cycle, or slowed down, resulting in energy savings but then speeding up will take time, as will be described below. Therefore, it is possible that in the holding stage, the pumping system consumes a much lower amount of energy due to the reduced speed without prejudicing the maintenance of the target pressure, so that energy efficiency is improved and pumping performance is maintained at the same time.

[0004] Therefore, considering the transitions between consecutive cycles involved in the above-mentioned cyclic operation, there is still a performance loss and energy waste. More specifically, the pump operating at the lowest speed in the holding phase will have to accelerate at the point in time when the next cycle begins in order to reach the maximum speed required for the pump-down phase in the next cycle. Therefore, the pumping system will have to wait for a delay as the pump accelerates from the lowest speed to the maximum speed. Due to this acceleration, there is no energy waste, but the overall pump-down time is increased. Longer pump-down times mean fewer batches per hour for the customer, resulting in less throughput.

[0005] Therefore, there is still a loss of pump performance and energy inefficiency when controlling the speed of the pump. Summary of the invention

[0006] It is therefore an object of the present invention to provide a pump having a speed control device which is able to have a higher energy efficiency and maintain the required satisfactory pump performance.

[0007] A solution to the above-mentioned problem is provided by a speed control method according to claim 1 , a speed control device according to claim 12 and a pump according to claim 17 .

[0008] The method for controlling the speed of a pump according to the invention is particularly suitable for variable speed drive (VSD) vacuum pumps employed in cyclic applications. Cyclic applications of pumps generally involve the pump operating in repeated cycles, wherein during each cycle the pump can operate at different speeds, which is made possible by the VSD. Therefore, the method comprises controlling the pump to operate in repeated cycles, and each cycle comprises an evacuation phase and a holding phase, so that the pump is controlled to repeatedly evacuate the volume in the vacuum chamber. For example, in each cycle, the pump is first operated at a higher speed in order to maintain a reduced pressure, and then, after reaching a certain desired target pressure, is operated at a lower speed sufficient to maintain the desired target pressure. The method analyzes the cycle duration based on the rise in the inlet pressure. The method takes into account the pump acceleration time and will calculate what is the best moment to accelerate again. This moment may be before the rise in pressure, thereby determining a new cycle. According to the method, the time elapsed between two consecutive cycles begins, if then the cycle time.

[0009] Furthermore, the method comprises determining the start of the next cycle during the current cycle, wherein the current cycle preferably directly precedes the next cycle. It is therefore possible to use the method to determine the start of the next cycle that is not yet coming, for example, directly after the current cycle, so as to ensure better control and calculation of the length of time involved at least in the holding phase.

[0010] Furthermore, the method comprises controlling the pump to accelerate to the maximum speed during the holding phase of the current cycle earlier before the start of the next cycle, which means that the time that would have been wasted in the holding phase of the current cycle is used for accelerating the pump, so that at the start of the next cycle the full pump capacity is already available for further and direct use in the pump-down phase of the next cycle.

[0011] The invention therefore has the advantage that, by accelerating the pump in advance, and more specifically at the end of the holding phase in the immediately preceding cycle, the acceleration delay that would have occurred at the beginning of each pump-down phase is avoided, so that energy that would have been wasted during the holding phase is also saved. In other words, by using the end of each holding phase to accelerate the pump, the energy efficiency of the entire pumping system is increased, and the performance loss caused during the pump-down phase is also reduced due to the elimination of the acceleration delay.

[0012] Preferably, during the evacuation phase, the pump is controlled to operate at maximum speed so that the volume in the vacuum chamber is evacuated from the high pressure to a set point or target pressure. The maximum permissible speed of the pump is preferably a function of the inlet pressure of the pump, the motor current of the pump and the oil temperature. The maximum permissible speed is therefore an inherent parameter of the operational requirements of the pumping system. In this case, it is ensured that the pump is controlled to evacuate the volume as quickly as possible in order to provide an optimal evacuation time, which, however, inevitably involves a maximum amount of energy consumption.

[0013] Preferably, during the holding phase, the pump is controlled to gradually slow down to a minimum or minimal speed such that the above set point pressure is maintained, thereby ensuring the overall performance of the pumping system. Thus, energy efficiency has been increased because the holding phase requires a much lower speed to maintain the set point pressure without compromising the performance of the pump.

[0014] More preferably, during the holding phase, the speed reduction of the pump is controlled based on the inlet pressure of the pump and the set point pressure. Thus, preferably, the pump speed is reduced linearly with the pump pressure, following a function determined by the inlet pressure and the set point pressure and preferably by further taking into account the cycle time of the ongoing cycle (i.e. the length of time between the start of the current cycle and the start of the next cycle).

[0015] Preferably, the method further comprises detecting the start of each cycle upon detecting that the inlet pressure reaches a preset threshold value. During the repeated cycles, the preset threshold value of the inlet pressure is preferably greater than the set point pressure and less than the maximum inlet pressure.

[0016] Preferably, when determining the start of the next cycle during the current cycle, a Bayesian filter, in particular a Kalman filter, is applied so that the start of the next cycle is estimated based on the start of at least one previous cycle, which preferably directly precedes the next cycle. In other words, given the start of at least one previous cycle before the next cycle, the method includes using the information contained in the start time point of the at least one previous cycle to estimate the start of the next cycle that has not yet arrived. Such an estimation can be achieved by a Bayesian filter, and in particular a Kalman filter.

[0017] Preferably, the inlet pressure of the pump is estimated by implementing a Kalman filter on the derivative of the inlet pressure, wherein the derivative of the inlet pressure is preferably performed over time. The result of the Kalman filter is then monitored over time in order to be compared with the preset threshold of the inlet pressure mentioned above to decide whether a new cycle starts. More specifically, it is preferred that the real-time value of the inlet pressure is monitored by the Kalman filter so that once the preset threshold is reached by the estimation of the inlet pressure, the start of a new cycle (i.e., preferably the next cycle that directly follows the current cycle) can be detected. In addition, not only the derivative of the inlet pressure can be estimated by the Kalman filter, but also the inlet pressure itself, as long as the rise in the inlet pressure or the absolute value of the inlet pressure can be estimated to determine the start of a new cycle. Therefore, the present invention is not intended to be limited to the selection of parameters estimated using a Bayesian filter and in particular a Kalman filter.

[0018] More specifically, the estimation of the inlet pressure is based on the following steps: based on an estimation of the derivative over time of the inlet pressure in at least one previous cycle, preferably directly before the next cycle, A priori Predicting the time derivative of the inlet pressure in the next cycle; and a priori Prediction and observation of the time derivative of the inlet pressure in the current cycle, A posteriori In other words, given information about the estimated value of the derivative of the inlet pressure over time in at least one previous cycle before the next cycle (i.e., the cycle whose information is to be estimated), the derivative of the inlet pressure over time in the next cycle is predicted. a priori information; and the predicted value needs to be further updated taking into account the observed values ​​that can be measured by the control method in the current cycle. Therefore, the estimate of the inlet pressure in each cycle is a combined result obtained from the estimate of at least one previous cycle and the observed or measured values ​​obtained in the current cycle, so that the accuracy of the estimate is ensured. As already mentioned above, the choice of specific parameters estimated by Kalman filtering is not limited to the derivative of the inlet pressure. In addition, preferably, the number of previous cycles considered is at least one, but more preferably two or three, and can be more than three, depending on the requirements of the estimation accuracy and the complexity of the algorithm.

[0019] Preferably, the method further comprises estimating the pre-acceleration time of the next cycle. It is necessary to estimate the time point during the next cycle at which the pump is to be controlled to accelerate, so that at the beginning of the next cycle, the pump is already at its full capacity. It is to be noted that since the beginning of a cycle can be determined when a preset threshold is reached, the full capacity of the pump is considered to be reached when a preset threshold of the inlet pressure is reached. Therefore, the pre-acceleration time associated with the next cycle refers to the length of time from the beginning of the next cycle until the time point in the holding phase of the next cycle when the pump is controlled to accelerate. Depending on the specific application, the beginning of the next cycle can be fixed or estimated as described above, and given the estimation of the pre-acceleration time of the next cycle, it is possible to control the pump to accelerate to its maximum speed in advance at the time point when the estimated pre-acceleration time expires in the holding phase. In addition, the estimation is implemented by a Bayesian filter and in particular a Kalman filter based on the pre-acceleration time of at least one previous cycle, preferably directly before the next cycle. The number of previous cycles considered is more preferably two or three, and at least one previous cycle does not have to be directly before the next cycle whose information is to be estimated.

[0020] Preferably, the estimate of the pre-acceleration time of the next cycle is obtained by a weighted average of an estimate of the pre-acceleration time of at least one previous cycle, preferably directly before the next cycle, and an observation of the pre-acceleration time of the current cycle. Advantageously, the information of the previous estimate and the observation of the current measurement are combined in order to obtain a better balance of the estimate of the information related to the next cycle.

[0021] Preferably, the following iterative learning formula is applied to estimate the pre-acceleration time of the next cycle,

[0022]

[0023] in:

[0024] j = iteration index for the current loop,

[0025] u = estimate of pre-acceleration time,

[0026] a = weighting factor,

[0027] b = weighting factor,

[0028] c = weighting factor,

[0029] d = weighting factor,

[0030] CycleTime = cycle time observation,

[0031] Acceleration = the acceleration constant in rpm / s used in the VSD inverter, and

[0032] offset=Offset parameter to increase stability.

[0033] In addition, although estimates of three previous cycles are applied in the above-mentioned iterative learning formula, depending on the specific requirements for estimation accuracy and the complexity of the algorithm, only one or two or more than three previous cycles may be applied to the formula, which naturally leads to a change in the number of weighting factors used in the formula. In addition, the weighting factors appear as fixed parameters in the above-mentioned formula, however, this can also be estimated for each cycle accordingly by using a Bayesian or Kalman filter. In addition, an offset parameter is applied to increase the stability of the above-mentioned iterative learning algorithm. Therefore, the offset parameter is a fixed value for a specific application scenario, or a value that can be selected by performing several tests on the above-mentioned formula before applying it. As shown in the above formula, the choice of offset depends on the difference between the estimated value of the pre-acceleration time of any previous cycle as mentioned above and the value obtained after further subtracting the time required for the pump to accelerate from its minimum or lowest speed to its maximum speed from the observation of the cycle time.

[0034] Preferably, when the above control method is run for the first time, the pump is controlled to first operate at the maximum speed for at least one cycle, so that the start time point and / or pre-acceleration time of at least one cycle is obtained, wherein the number of cycles during which the pump is controlled to operate at the maximum speed depends on the number of previous cycles to be considered for the estimation as described above. That is, as long as sufficient values ​​of the iteration term suitable for the first estimation are obtained, the pump is controlled to operate at the maximum speed.

[0035] The device for controlling the speed of a pump according to the present invention is particularly suitable for VSD vacuum pumps used in cyclic applications. The device comprises a memory unit in which instructions regarding the steps of the method as mentioned above are stored, and in which preferably parameters fixed or updated during the estimation as mentioned above are stored. For example, a first sub-memory unit is preferably provided, which includes an iteration index and an estimate of the pre-acceleration time and the cycle time of at least one previous cycle. The second sub-memory unit is used to store fixed values ​​of parameters, such as weighting factors, acceleration, maximum speed and minimum speed of the pump, and offset parameters. The speed control device further comprises a processing unit, which is configured to execute the instructions included in the memory unit; an output unit, which is configured to transmit a speed control signal to the pump; and an input unit, which is configured to read the inlet pressure of the pump and feed it to the processing unit.

[0036] Preferably, the apparatus comprises a time counting unit configured to calculate a starting time point and / or a pre-acceleration time of at least one cycle when the processing unit starts to execute the instruction for the first time.

[0037] Preferably, the time counting unit is further configured to measure the start time point and / or the cycle time of the current cycle. Thus, during the current cycle, the estimation of the next cycle takes into account the actual values ​​measured in the current cycle, resulting in a higher accuracy and better balance of the results of the estimation performed in the current cycle.

[0038] There is also provided a computer program according to the invention, the computer program comprising instructions for causing a speed control device as mentioned above to perform the steps of the method as mentioned above.

[0039] There is also provided a computer readable medium according to the present invention, in which the above-mentioned computer program is stored.

[0040] The pump according to the present invention is particularly a VSD vacuum pump used in non-circulatory applications, which is coupled to a device for controlling its speed as mentioned above. Wherein, the vacuum pump can be a dry claw pump, a roots pump, a vortex pump or a rotary screw pump, and a variable speed drive can be coupled to the pump so as to be suitable for specific application requirements. In addition, the vacuum pump can be a volumetric pump, such as a diaphragm pump, a gas or steam volumetric pump and a piston pump. In addition, the vacuum pump can also be a speed pump, such as an inertial pump. The pump includes a control unit, which is configured to detect multiple pump parameters, wherein preferably, multiple sensors are provided in the pump for detecting parameters by the control unit. The pump further includes an output unit, which is configured to transmit multiple pump parameters to a speed control device coupled to the pump. In addition, an input unit is included in the pump for receiving a speed control signal transmitted from the speed control device so that the pump is controlled to operate according to the steps of the method as mentioned above. If the pumping system and the speed control device need to be physically located at one site, the communication between the input / output unit of the pump and the speed control device can be achieved by wires. However, the speed control device may also be located in another location than the pump, in which case wireless communication is applied between the speed control device and the pump. Furthermore, it is also possible to provide a central control device at another location, which relays or controls the signal transmission between the pump and the speed control device, so that several pumps at different locations can be controlled simultaneously. Thus, according to the invention, both centralized / decentralized and local / networked arrangements of the pumping control system can be achieved.

[0041] Preferably, at least the inlet pressure of the pump is included among the plurality of pump parameters transmitted from the control unit of the pump to a pump speed control device or a control centre located locally at the pump location or remote from the pump station.

[0042] Thus, when applied to cyclic applications in industry, the invention has the advantage, compared to conventional pumps without variable speed drives or pumps with variable speed drives but without specific speed control of the pump during each cycle, that each repetitive cycle during the application of the pump is divided into two phases, namely a pump-down phase and a holding phase; that in the pump-down phase, the pump is controlled to pump down the volume as quickly as possible by operating at maximum speed in order to provide the best pump-down time performance; that in the holding phase, the pump is controlled to have already started to accelerate to its full capacity, thereby utilizing the energy during this holding phase to compensate for the acceleration delay that would otherwise occur at the beginning of the pump-down phase of the next cycle; and that Bayesian filtering, and more particularly Kalman filtering, is utilized to accurately estimate the start time point of each cycle and the length of time between the start of the cycle and the point in time at which the pump is controlled to accelerate during the holding phase of the cycle by taking into account the estimated information obtained for the previous cycle and the observed information obtained during the current cycle. Thus, the speed of the pump is continuously adapted to the real-time operating conditions of the pump, resulting in avoiding energy waste during the holding phase and avoiding a loss of pump performance during the pump-down phase. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In the following disclosure, reference will be made to the accompanying drawings, in which:

[0044] Figure 1 is a graph illustrating the relationship between pump speed and pressure,

[0045] Figure 2 is a graph illustrating the start of a repetitive cycle and the acceleration of the pump during a holding phase, each cycle comprising a pump-down phase and a holding phase,

[0046] Figure 3 is a graph illustrating an estimate of the start of the next cycle and an estimate of the point in time of acceleration of the pump during the next cycle,

[0047] Figure 4 is a flow chart illustrating the iterative speed control process,

[0048] Figure 5 is a schematic representation of a speed control device,

[0049] Figure 6 is a schematic representation of the pump, and

[0050] Figure 7 is a schematic illustration of a pumping system involving a speed control center coupled to a plurality of pumps. DETAILED DESCRIPTION

[0051] Figure 1The relationship between pump speed and pressure is illustrated, with the maximum speed of the pump shown as 6000 revolutions per minute (rpm), the minimum speed of the pump as 3500 rpm, the set point pressure or desired target pressure as 50 millibars (mbar), and the hysteresis shown as extending over 30 mbar. Figure 1 , the pump pressure starts at a value as high as 1000 mbar, while the pump is operated at maximum speed so that the pressure continues to be evacuated. At the point in time when the set point pressure of 50 mbar is reached, the pump gradually slows down, which leads to a hysteresis area, during which the pressure also gradually decreases until it reaches 20 mbar. Thereafter, the pump is operated at minimum speed, which leads to a further decrease in pressure. Because in the later stage after the set point pressure of the pump is reached, the pump is operated at a much lower speed while still maintaining satisfactory pump pressure performance, a lot of energy is saved.

[0052] refer to Figure 2 , presents a detailed representation of a repetitive cycle, each cycle comprising a pump-down phase and a holding phase, wherein the start of the repetitive cycle and the point in time during the holding phase when the control pump is accelerated are also shown. Figure 2 In the embodiment, the current cycle extends between the start of the current cycle and the start of the next cycle, wherein the current cycle starts at time point 1 TP1, when the rise in the inlet pressure of the pump is considered to be large enough, in this case, when the inlet pressure reaches 700 mbar, and the start of the next cycle (i.e. time point 4 TP4) is determined similarly to the current cycle. Although it is common practice that the determination of the start of a cycle relies on a fixed value of the inlet pressure increment or an absolute value of the inlet pressure, this criterion can also be changed for each cycle during a cycle application and can further depend on the specific application scenario. Figure 2 Each cycle shown in Figure 1 consists of two phases, namely the pump-down phase and the holding phase. The pumping process enters the holding phase at time point 2 TP2, when the set point pressure is reached, which in this case is 50 mbar (this value is the same as Figure 1 ). Figure 2 , at time point 3TP3, the pump is controlled to increase its speed before the start of the next cycle (i.e. TP4), so that shortly after TP4, the full pump capacity is already available. In other words, by increasing the speed of the pump at the holding phase of the current cycle, the remaining time of the holding phase can be used to compensate for the acceleration delay, which would otherwise occur after the start of the next cycle, resulting in a performance loss of the pumping system. According to the present invention, the start of the next cycle is determined during the current cycle in a manner similar to the way the start of the current cycle is determined.

[0053] Despite Figure 2It is not directly depicted in FIG, but it is preferred that during the evacuation phase of each cycle, the pump is controlled to operate at maximum speed so as to evacuate the volume from the high pressure to the set point pressure, wherein Figure 2 In the case of , the high pressure is shown as 1000 mbar and the set point pressure is shown as 50 mbar. In addition, it is preferred that during the holding phase of each cycle, the pressure is gradually reduced at a much slower rate than during the evacuation phase. Figure 2 , during the holding phase of the current cycle, the pump pressure gradually approaches 0 mbar. According to the present invention, even more preferably, during the holding phase, the speed of the pump is controlled to decrease based on the inlet pressure and the set point pressure in order to achieve better control of the holding phase.

[0054] Therefore, at least according to Figure 2 Obviously, predetermining the point in time when the pump will accelerate during the holding phase of the current cycle and determining the start of the next cycle ensures that plenty of time is reserved for the pump to accelerate to its full capacity at the start of the next cycle.

[0055] refer to Figure 3 , presents a more detailed graph illustrating an estimate of the cycle start and time point of pump acceleration, where the overall pumping process during each cycle is similar to that for Figure 2 According to the content already described. Figure 3 , the next cycle C2 starts at the time point TP4, which is determined according to the invention based on the start time point of at least one previous cycle, which in the present case refers to at least one of the cycles C1, C0 and the cycle before C0 before the next cycle C2. Preferably, the estimation of the cycle start of C2 is achieved by a Bayesian filter, and more particularly and preferably a Kalman filter. Similarly, at least one previous cycle before C3 is used to determine the cycle after the next cycle C2, i.e. C3 ( Figure 3 The estimation of the cycle start is therefore performed repeatedly for any new cycle using information from the previous cycle before the new cycle. However, depending on the application scenario and requirements, the cycle estimation may be implemented only for some cycles during the entire pumping process.

[0056] According to Figure 3 In the invention illustrated in FIG. 1 , the time point TP4 is estimated by preferably applying a Kalman filter to the derivative of the inlet pressure over time. More specifically, during the current cycle C1, information on the start of the cycles C1, C0 and the cycle before C0 is already at hand, wherein information on at least one of these cycles C1, C0 and the cycle before C0 is used to predict the time point TP4. a prioriinformation; moreover, during the current cycle, the actual value of the time derivative of the inlet pressure can already be observed and measured, which is therefore used to further update the prediction information of TP4, resulting in a more accurate prediction of TP4. Posterior estimate.

[0057] In summary, each of TP1, TP4, and TP7 (the start of the cycle of C3 at time point 7) is similarly estimated in the manner described above by using both the previous estimate of the relevant parameter (i.e., the start of the cycle of at least one corresponding previous cycle) and the observation or measurement information of one of the previous cycles. Note that the number of previous cycles applied in obtaining the estimate is at least one and depends on the specific application scenario and requirements; and although in Figure 3 In the example, an observation using the cycle start of the current cycle directly before the next cycle appears, but depending on the application scenario and requirements, another or more than one measured value or observed value of the relevant parameter can also be applied.

[0058] Further based on Figure 3 Even more preferably, for each cycle, the length of time between the start of the cycle and the point in time when the pump is controlled to accelerate is also estimated. Figure 3 It is shown in FIG. 1 that during the current cycle C1, time point 3 TP3 is the time point when the pump needs to accelerate, which has been determined during the execution of the previous cycle (i.e., cycle C0 and the cycle before C0), and similarly, for the next cycle C2, time point 6 TP6 is also determined during the previous cycle C1, C0 and the cycle before C0. Since according to the present invention, the start of each cycle is a fixed value or estimated as described above, Figure 3 The estimation of the pre-acceleration time illustrated in gives an estimation of the point in time when the pump needs to be controlled to accelerate. Preferably, according to the invention, the estimation of the acceleration time point is also performed by a Bayesian filter, and more particularly and preferably, by a Kalman filter based on information of at least one of the pre-acceleration times of the previous cycles C1, C0 and the cycle before C0.

[0059] Apart from Figure 3 In addition, and according to the present invention, the estimation of the pre-acceleration time preferably also includes two steps, one of which is to predict the pre-acceleration time for the next cycle C2 based on at least one estimation of the previous cycle C1. a priori information, i.e., C0 and the cycle before C0, and another is to update the prediction using observation or measurement information about one of the previous cycles C1, C2 and the cycle before C0, and wherein preferably, the measurement information is obtained from the cycle (i.e., C1) directly before the next cycle C2.

[0060] According to the invention, it is preferred that a weighted average of an estimate of the pre-acceleration time of at least one of the previous cycles C1, C0 and the cycle before C0 and an observation of one of the previous cycles C1, C0 and the cycle before C0 is applied to the estimate. According to the invention and more preferably, the weighted average is in the following form of an iterative learning formula,

[0061]

[0062] in:

[0063] j = iteration index for the current loop,

[0064] u = estimate of pre-acceleration time,

[0065] a = weighting factor,

[0066] b = weighting factor,

[0067] c = weighting factor,

[0068] d = weighting factor,

[0069] CycleTime = cycle time observation,

[0070] Acceleration = the acceleration constant in rpm / s used in the VSD drive, and

[0071] offset=Offset parameter to increase stability.

[0072] In the above formula, it is assumed that the current loop has iteration index j and three previous loops directly before the next loop denoted by j+1, denoted by j, j-1 and j-2 respectively, are applied. Depending on the application scenario and requirements, it is also possible to use less than three previous loops. As shown in the above formula, an additional balancing term is included, which involves at least the observation of the loop time of the current loop. Reference Figure 3 , the observed value of the pre-acceleration time is equal to the observed value of the cycle time minus the time required for the pump to accelerate from the minimum speed to a preset threshold, and this subtraction result may therefore be involved in the balancing term. Preferably, since the preset threshold is large enough and close to the maximum speed of the pump, the balancing term may instead include the observation of the cycle time minus the time required for the pump to accelerate from the minimum speed to the maximum speed. It is further preferred that an offset parameter is inserted into the balancing term to increase the stability of the iterative learning algorithm, and the offset parameter may be fixed or adapted during each cycle of the iterative method according to the application scenario and requirements. In addition, the weighting factors a, b, c and d are shown as fixed parameters in the above formulas, and according to the present invention, it is also possible to update these weighting factors in a similar manner to the pre-acceleration time.

[0073] Figure 4 is a flow chart depicting a speed control method according to the present invention. Therein, the speed control method starts at step 10, i.e., S10, for initialization of parameters. Since information of at least one previous cycle will be used for estimation of the next cycle, it is necessary to initialize the parameters when the speed control is activated for the first time. More specifically, a safe mode is entered, in which the pump is controlled to run at its maximum speed for a plurality of cycles in order to obtain the parameters. For example, as shown in the above iterative learning formula, the pump can be controlled to operate in the safe mode for three cycles, based on which an estimation of the fourth cycle can be performed, and the estimation can be repeated for each cycle after the fourth cycle.

[0074] At step 20, i.e. S20, it is tracked whether a new cycle has started. The criterion used here is a sufficiently large rise in the inlet pressure of the pump, or the absolute value of the inlet pressure reaches a preset threshold according to the present invention. Therefore, by comparing the inlet pressure measured at any time point with the preset threshold, the start of any new cycle at this step can be simply detected, and the preset threshold can be a fixed value or an unfixed value according to the application scenario and requirements. Preferably, it is also possible to estimate the inlet pressure of the new cycle at this step by, for example, implementing a Kalman filter as mentioned above. Taking the fourth cycle as mentioned above as an example, the derivative of the inlet pressure at any time point is updated in real time, so that when the estimation result reveals that the inlet pressure is higher than the preset value of the inlet pressure given the previous information of the cycle start of the first, second and third cycles, the time point is determined to be the start of the fourth cycle. If it is confirmed that a new cycle has just started after checking S20, the speed control method proceeds to step 30, i.e. S30, where the cycle time (which is the time between the start of two cycles) is then counted so that all iteration parameters are reset. Referring back to the learning control formula shown above, during each iteration, it is necessary to update the three pre-acceleration times used therein, namely u j 、u j-1 and u j-2 , and the observation of cycle time, i.e., CycleTime j If necessary, it is also possible to reset the parameters of the weighting factors and the offset at S30. On the other hand, if after checking S20 it is confirmed that the pump is operating under conditions other than the start of a cycle, the speed control method proceeds to step 40, ie S40.

[0075] At S40, it is checked whether the inlet pressure is not higher than the set point pressure Psp, or whether the pump is operating in a safe mode. If neither criterion is met, the speed control method proceeds to step 50, i.e. S50, where the pump is controlled to remain operating at the maximum speed. In other words, the pump is still used to operate in the pump-down phase, where the inlet pressure has not yet dropped to the set point pressure.

[0076] On the other hand, if at S40, it is determined that the inlet pressure has dropped to or is below the set point pressure, or the pump is operating in safety mode. The speed control method then proceeds to step 42, i.e., S42, in order to check whether the iterative learning control ILC is activated for estimating the pre-acceleration time of the following cycle. Therefore, it is possible that at S42, the result is that the ILC is not activated, thereby indicating normal operation of the pump, which is illustrated by step 52 (i.e., S52). At S52, the speed of the pump is controlled to slow down based on the inlet pressure and the set point pressure Psp. It is preferred to consider the time and the cycle time to determine the manner in which the speed is to be reduced.

[0077] On the other hand, at S42, when it is confirmed that the ILC is activated, the speed control method then proceeds to step 44, i.e., S44, where it is checked whether the pre-acceleration time has expired, which means that the pump needs to be controlled to accelerate again. Taking the fourth cycle as mentioned above as an example, preferably given the relevant information of the first three cycles, an estimate of the start of the fourth cycle and the pre-acceleration time required in the fourth cycle have preferably been obtained during the third cycle. When the pre-acceleration time calculated from the start time point of the fourth cycle expires, it is confirmed at S44 that the speed control method can then proceed to step 56, i.e., S56, where the pump is controlled to run at maximum speed. In other words, after the pre-acceleration time expires, before the start of the next cycle that has not yet arrived, the pump has accelerated in the holding phase of the current cycle.

[0078] On the other hand, if it is confirmed at S44 that the pre-acceleration time has not expired, this means that the pump only needs to continue to slow down, similar to what is described in S52 .

[0079] Furthermore, steps S50 , S52 , S54 and S56 are further advanced to step 60 , ie, S60 , to a speed limiter so that the maximum allowable speed of the pumping system is not exceeded.

[0080] Figure 5The speed control device 10 according to the present invention is illustrated, wherein the speed control device 10 comprises at least a memory unit 100, a processing unit 102, an output unit 104 and an input unit 106. The memory unit 100 comprises at least instructions regarding the method steps as mentioned above, which are executed by the processing unit 102. It is also possible that relevant parameters, including preset parameters, such as preset thresholds for inlet pressure, weighting factors used in iterative learning control formulas, and offsets (if they are constant values), are stored in a subsection included in the memory unit 100. Therefore, it is also possible to have another subsection in the memory unit 100 to store all iteration indexes and all relevant parameters that will change during the iteration, including an estimate of the start time point of the cycle and an observation of the cycle time. In addition, the maximum speed, minimum speed and acceleration of the pump are pump-specific parameters that are fixed according to the application scenario and requirements and can therefore be stored in another storage subsection included in the memory unit 100. The output unit 104 transmits the speed control signal given by the speed control device 10 and parameters including at least the inlet pressure of the pump to be read from the input unit 106 to the pump.

[0081] Also like Figure 5 As shown in FIG. 1 , it is preferred that a time counting unit 108 is included in the speed control device 10 , wherein the time counting unit 108 is particularly configured to mark the time point of the start of the cycle and calculate the pre-acceleration time of the cycle, which is used to keep track of the start time point of the cycle, such as Figure 4 As shown in S30 in FIG. 1 , the pre-acceleration time is counted to see whether it has expired, such as Figure 4 In addition, it is preferred that the time counting unit 108 measures an observation of a period of time whenever it is needed as a balancing term in the estimation method, for example, the observation or measurement of the pre-acceleration time of the current cycle.

[0082] Figure 6 A pump 20 according to the invention is depicted, which comprises at least a control unit 202, an input unit 204 and an output unit 206. The output unit 206 transmits at least one pump parameter including the pump inlet pressure to a speed control device, such as Figure 5 The speed control device 10 shown in FIG. 1 is a speed control device 10. The input unit 204 is configured to receive a signal from a speed control device (such as Figure 5 The pump 20 may further include at least one sensor in order to obtain the pump parameters required by the control unit 202 .

[0083] refer to Figure 7And according to the present invention, a system view of a pumping control system including the above-mentioned speed control device is presented. More specifically, the pumping system includes a speed control center 12, which controls a plurality of pumps or pumping systems 62, 64 and 66, such as Figure 7 . According to the present invention, the output unit and the input unit of the speed control center 12 are connected to the input units and the output units of the plurality of pumps 62, 64 and 66, respectively. The speed control center 12 may be located at the same location as at least one of the plurality of pumps or pumping systems, in which case at least one of the connections 602, 604 and 606 between the output / input unit of the speed control center 12 and the plurality of pumps is a physical cable having a transmission rate capable of handling the transmission between them according to the application scenario and requirements. Therefore, it is also possible that the speed control center 12 is located in another location different from one or more or all of the pumps or pumping systems, in which case one or more or all of the connections between them are realized by wireless communication. Therefore, according to the present invention, not only a centralized local control network of pump speeds centered on the speed control center 12 is possible, but also a decentralized pump speed control system is achievable, in which the speed control center 12 is capable of controlling at least one pump or pumping system simultaneously, resulting in even higher efficiency of the control method, and a reduction in the demand for physical devices for speed control.

[0084] The invention is in no way limited to the embodiments described as examples and shown in the drawings, but such methods and devices for controlling the speed of a pump, computer programs and computer-readable media on which the computer programs applied thereby are stored, and pumps may be implemented in all kinds of forms without departing from the scope of the invention.

Claims

1. A method for controlling the speed of a pump, the method comprising: controlling the pump to operate in repeated cycles, each cycle comprising an evacuation phase and a holding phase, wherein each cycle starts at a point in time when the rise in the inlet pressure of the pump is sufficiently large, and the time between the start of two consecutive cycles is the cycle time; During a current loop, determining the start of a next loop, wherein the current loop directly precedes the next loop; and prior to the start of a next cycle, controlling the pump to accelerate to a maximum speed during a holding phase of a current cycle so that at the start of the next cycle, full pump capacity is available; During the hold phase, the pump is controlled to slow down to a minimum speed such that the set point pressure is maintained.

2. The method according to claim 1, wherein During the evacuation phase, the pump is controlled to operate at maximum speed so that the volume is evacuated from the high pressure to the set point pressure.

3. The method according to claim 1 or 2, wherein The pump is a variable speed drive VSD vacuum pump used in a circulation application.

4. The method according to claim 3, wherein During the hold phase, the pump speed is controlled to decrease based on the inlet pressure and the set point pressure.

5. The method according to claim 1 or 2, wherein the method further comprises The start of each cycle is detected when the inlet pressure reaches a preset threshold.

6. The method according to claim 1 or 2, wherein The start of the next cycle is estimated by a Bayesian filter based on the start of at least one previous cycle directly before the next cycle.

7. The method according to claim 6, wherein The Bayesian filter is a Kalman filter.

8. The method according to claim 6, wherein The inlet pressure is estimated by applying a Kalman filter to the time derivative of the inlet pressure based on the following steps: predicting a priori the derivative of the inlet pressure with respect to time in the next cycle based on an estimate of the derivative of the inlet pressure with respect to time in at least one previous cycle directly before the next cycle; and Based on the a priori prediction of the time derivative of the inlet pressure in the next cycle and the observation of the time derivative of the inlet pressure in the current cycle, the time derivative of the inlet pressure in the next cycle is updated a posteriori, where By observation, it is called a measurement, and by estimation, it is called an updated prediction based on the observation.

9. The method according to claim 1 or 2, wherein: The method further includes estimating a pre-acceleration time of a next cycle, the pre-acceleration time referring to the length of time from the start of the next cycle until a time point in the holding phase of the next cycle when the pump is controlled to accelerate, wherein The estimation is performed by a Bayesian filter based on the pre-acceleration time of at least one previous cycle directly before the next cycle.

10. The method according to claim 9, wherein The Bayesian filter is a Kalman filter.

11. The method according to claim 9, wherein The estimate of the pre-acceleration time for the next cycle is a weighted average of the estimate of the pre-acceleration time for at least one previous cycle directly before the next cycle and the observation of the pre-acceleration time for the current cycle.

12. The method according to claim 11, wherein the method further comprises Estimate the pre-acceleration time of the next cycle based on the iterative learning formula, in: j = iteration index for the current loop, u = estimate of pre-acceleration time, a = weighting factor, b = weighting factor, c = weighting factor, d = weighting factor, CycleTime = observation of cycle time, Acceleration = the acceleration constant in rpm / s used in the VSD drive, and offset = offset parameter used to increase stability.

13. The method according to claim 1 or 2, wherein: The method further comprises, when the control is run for the first time, controlling the pump to operate at a maximum speed for at least one cycle so that a start time point and / or a pre-acceleration time of the at least one cycle is obtained.

14. A device for controlling the speed of a pump, the pump being a variable speed drive (VSD) vacuum pump used in a circulation application, the device comprising a memory unit comprising instructions concerning the steps of the method of any one of claims 1 to 13; a processing unit configured to execute instructions included in the memory unit; an output unit configured to transmit a speed control signal to a pump; as well as An input unit configured to read the inlet pressure of the pump and feed it to the processing unit.

15. The device according to claim 14, wherein the device further comprises A time counting unit is configured to calculate a start time point and / or a pre-acceleration time of at least one cycle when the processing unit starts to execute instructions for the first time.

16. The device according to claim 14, wherein The time counting unit is further configured to measure a start time point of a current cycle and / or a cycle time.

17. A computer program comprising instructions for causing the speed control device of any one of claims 14 to 16 to execute the steps of the method of any one of claims 1 to 13.

18. A computer readable medium having stored thereon the computer program of claim 17.

19. A pump coupled to a device for controlling the speed of the pump, the pump comprising a control unit configured to detect a plurality of pump parameters; an output unit configured to transmit a plurality of pump parameters to a speed control device; as well as An input unit configured to receive a speed control signal transmitted from a speed control device so that the pump is controlled to operate according to any one of claims 1 to 13.

20. The pump of claim 19, wherein The plurality of pump parameters includes the inlet pressure of the pump.

21. The pump of claim 19, wherein the pump is a variable speed drive (VSD) vacuum pump employed in a circulation application.

Images