Method and device for identifying use mode of forklift scale and computer readable medium

Abstract

The invention provides a use mode identification method and device of a forklift scale and a computer readable medium. The method comprises the steps that using modes of the forklift scale are defined, the using modes at least comprise a static mode and a moving mode, and the moving mode comprises a dragging mode and / or a lifting mode; first data in each use mode is collected, the first data comes from an acceleration sensor arranged on the forklift scale, and the forklift scale is in at least one use mode at the same time; preprocessing the first data to obtain second data; performing feature value extraction processing on the second data to obtain a feature value; training a classifier according to the feature values and the corresponding use modes; and using a trained classifier to identify the use mode of the forklift scale according to the current characteristic value of the forklift scale. The identification method is high in accuracy and low in cost.

Description

Technical Field

[0001] The present invention mainly relates to the field of movable weighing instruments, and in particular to a method, a device and a computer-readable medium for recognizing a usage mode of a forklift scale. Background Art

[0002] In the pharmaceutical, chemical, food, auto parts and other industries, many manufacturers use pallet truck scales with instrument components to handle large quantities of cargo transportation. This pallet truck scale can not only ship cargo, but also weigh cargo anytime and anywhere, without the need to pull the cargo to the traditional platform scale for weighing. In some scenarios where the weight of the stacked cargo needs to be strictly managed, the forklift scale can be used to place the cargo directly according to the weighing results to meet the cargo placement needs, which is convenient and fast.

[0003] During the use of forklift scales, different working modes or states such as forklift stationary, gear lifting and lowering, and load dragging are involved. The weighing results of forklift scales are often affected by the working mode. Therefore, the weighing results obtained in different working modes may be different, and their accuracy is also different. With the increasing requirements for weighing accuracy and the need for trade settlement, it is necessary to further improve the weighing accuracy of forklift scales. Therefore, it is necessary to identify the working mode of forklift scales in order to obtain dynamic weighing results related to the working mode of forklift scales. However, the current examples of applying pattern recognition methods to specific products usually consider the following two aspects: (1) In order to improve the recognition accuracy, it is necessary to use more complex sensors to obtain sufficient measurement data, such as nine-axis inertial sensors including magnetometers, gyroscopes, and accelerometers; (2) In order to improve efficiency, simple feature values ​​are selected for recognition, and the categories that can be recognized are relatively small, such as only normal and abnormal. For aspect (1), complex sensors such as nine-axis inertial sensors are powerful and expensive. If they are used in the recognition of working modes of forklift scales, it will cause a waste of sensor functions. Regarding the second aspect, the working modes of forklift scales are relatively complex and diverse, and it is not enough to simply identify normal and abnormal conditions. Therefore, it is urgent to propose a method suitable for identifying the working modes of forklift scales. Summary of the invention

[0004] The technical problem to be solved by the present invention is to provide a recognition method, device and computer-readable medium for recognizing the working mode of a forklift scale based on an acceleration sensor.

[0005] To solve the above technical problems, the present invention provides a method for identifying a usage mode of a forklift scale, comprising: defining a usage mode of the forklift scale, the usage mode comprising at least a static mode and a motion mode, the motion mode comprising a dragging mode and / or a lifting mode; collecting first data under each of the usage modes, the first data coming from an acceleration sensor provided on the forklift scale, wherein the forklift scale is in at least one usage mode at the same time; preprocessing the first data to obtain second data; performing feature value extraction processing on the second data to obtain feature values; training a classifier according to the feature values ​​and corresponding usage modes; and using the trained classifier to identify the usage mode of the forklift scale according to the current feature values ​​of the forklift scale.

[0006] In one embodiment of the present application, a weighing sensor is also provided on the forklift scale, and the weighing sensor is used to obtain the weighing data of the forklift scale. The usage mode also includes a no-load mode and a load mode. When the weighing data is within a preset weight range, the usage mode is the no-load mode, otherwise, the usage mode is the load mode.

[0007] In an embodiment of the present application, the movement mode also includes a descending mode and / or a dragging and lifting mode.

[0008] In one embodiment of the present application, the step of preprocessing the first data includes: removing the first data exceeding a first threshold range as outliers.

[0009] In one embodiment of the present application, the first threshold range is μ±3δ in a normal distribution, where μ is the mean and δ is the standard deviation.

[0010] In one embodiment of the present application, the step of preprocessing the first data further includes:

[0011] The following formula is used to supplement the missing data after removing outliers:

[0012]

[0013] Among them, X miss is the missing data that needs to be supplemented, K is the total amount of data in the current group, and X miss-i and X miss+i X miss The previous set of data and the next set of data.

[0014] In one embodiment of the present application, the classifier includes a support vector machine classifier based on wavelet transform, and the step of preprocessing the first data also includes: determining the length of the time window corresponding to each of the usage modes and the number of continuous moving time windows taken.

[0015] In one embodiment of the present application, the acceleration sensor includes a three-axis acceleration sensor, the second data includes three-axis acceleration, and the step of performing eigenvalue extraction processing on the second data includes: performing a multi-resolution wavelet transform on the three-axis acceleration to obtain wavelet coefficients, the wavelet coefficients including high-frequency detail coefficients and low-frequency approximation coefficients; taking the sum of the squares of the wavelet coefficients as wavelet energy; selecting the peak value of the low-frequency approximation coefficient corresponding to a predetermined frequency range as the wavelet peak; the eigenvalue includes the wavelet energy and the wavelet peak.

[0016] In one embodiment of the present application, a result generated by combining the wavelet energy with the interquartile range is used as a eigenvalue corresponding to the wavelet energy.

[0017] In one embodiment of the present application, the step of taking the sum of the squares of the wavelet coefficients as the wavelet energy includes: taking the sum of the squares of the high-frequency detail coefficients of the third to seventh layers and the sum of the squares of the low-frequency approximation coefficients of the seventh layer as the wavelet energy; the step of selecting the peak value of the low-frequency approximation coefficient corresponding to a predetermined frequency range as the wavelet peak includes: selecting the peak value of the low-frequency approximation coefficient of the seventh layer as the wavelet peak.

[0018] In one embodiment of the present application, a result generated by combining the wavelet peak with the interquartile range is used as a characteristic value corresponding to the wavelet peak.

[0019] In one embodiment of the present application, the step of performing feature value extraction processing on the second data further includes: after obtaining the feature value, using a Wrapper feature subset selection method to perform dimensionality reduction processing on the feature value.

[0020] In order to solve the above technical problems, the present application also proposes a forklift scale usage pattern recognition device, comprising: a memory for storing instructions executable by a processor; and a processor for executing the instructions to implement the above method.

[0021] In order to solve the above technical problem, the present application further proposes a computer-readable medium storing a computer program code, wherein the computer program code implements the above method when executed by a processor.

[0022] The usage pattern recognition method of the forklift scale of the present application performs feature extraction on the first data obtained by the acceleration sensor, and extracts specific feature values ​​to train a classifier for the usage pattern of the forklift scale. The trained classifier can effectively classify the usage pattern of the forklift scale. The recognition device using the usage pattern recognition method of the present application is simple and only needs to set up a three-axis acceleration sensor to achieve accurate classification of the usage pattern without using other complex and expensive sensors. BRIEF DESCRIPTION OF THE DRAWINGS

[0023] The accompanying drawings are included to provide a further understanding of the present application. They are included and constitute a part of the present application. The accompanying drawings illustrate embodiments of the present application and together with the present specification serve to explain the principles of the present invention. In the accompanying drawings:

[0024] Figure 1 is a schematic diagram of the side view structure of a forklift scale applicable to the usage pattern recognition method of an embodiment of the present application;

[0025] Figure 2 is an exemplary flow chart of a method for using a pattern recognition according to an embodiment of the present application;

[0026] Figure 3 is a schematic diagram of various usage modes that can be identified by the usage mode identification method according to an embodiment of the present application;

[0027] Figure 4 is an exemplary flow chart of a method for identifying a usage pattern of a forklift scale according to an embodiment of the present application;

[0028] Figure 5A It is the result obtained by using the time domain features shown in Table 1 for classification;

[0029] Figure 5B It is the result obtained by using the frequency domain features shown in Table 2 for classification;

[0030] Figure 5C It is the result obtained by classification using the pattern recognition method of the present application;

[0031] Figure 6 It is a system block diagram of a usage pattern recognition device for a forklift scale according to an embodiment of the present application. DETAILED DESCRIPTION

[0032] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the following is a brief introduction to the drawings required for the description of the embodiments. Obviously, the drawings described below are only some examples or embodiments of the present application. For ordinary technicians in this field, the present application can also be applied to other similar scenarios based on these drawings without creative work. Unless it is obvious from the language environment or otherwise explained, the same reference numerals in the figures represent the same structure or operation.

[0033] As shown in this application and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0034] Unless otherwise specifically stated, the relative arrangement, numerical expressions and numerical values ​​of the parts and steps set forth in these embodiments do not limit the scope of the present application. At the same time, it should be understood that, for ease of description, the sizes of the various parts shown in the accompanying drawings are not drawn according to the actual proportional relationship. The technology, method and equipment known to those of ordinary skill in the relevant field may not be discussed in detail, but in appropriate cases, the technology, method and equipment should be considered as a part of the authorization specification. In all examples shown and discussed here, any specific value should be interpreted as being merely exemplary, rather than as a limitation. Therefore, other examples of exemplary embodiments may have different values. It should be noted that similar reference numerals and letters represent similar items in the following drawings, so that once a certain item is defined in an accompanying drawing, it does not need to be further discussed in subsequent drawings.

[0035] In the description of the present application, it should be understood that the directions or positional relationships indicated by directional words such as "front, back, up, down, left, right", "lateral, vertical, perpendicular, horizontal" and "top, bottom" are usually based on the directions or positional relationships shown in the drawings, and are only for the convenience of describing the present application and simplifying the description. Unless otherwise specified, these directional words do not indicate or imply that the device or element referred to must have a specific direction or be constructed and operated in a specific direction, and therefore cannot be understood as limiting the scope of protection of the present application; the directional words "inside and outside" refer to the inside and outside relative to the contours of each component itself.

[0036] For ease of description, spatially relative terms such as "above", "above", "on the upper surface of", "above", etc. may be used here to describe the spatial positional relationship between a device or feature and other devices or features as shown in the figure. It should be understood that spatially relative terms are intended to include different orientations of the device in use or operation in addition to the orientation described in the figure. For example, if the device in the accompanying drawings is inverted, the device described as "above other devices or structures" or "above other devices or structures" will be positioned as "below other devices or structures" or "below other devices or structures". Thus, the exemplary term "above" can include both "above" and "below". The device can also be positioned in other different ways (rotated 90 degrees or in other orientations), and the spatially relative descriptions used here are interpreted accordingly.

[0037] In addition, it should be noted that the use of words such as "first" and "second" to define components is only for the convenience of distinguishing the corresponding components. If not otherwise stated, the above words have no special meaning and cannot be understood as limiting the scope of protection of this application. In addition, although the terms used in this application are selected from well-known and commonly used terms, some terms mentioned in the specification of this application may be selected by the applicant at his or her discretion, and their detailed meanings are explained in the relevant parts of the description of this article. In addition, it is required to understand this application not only by the actual terms used, but also by the meaning implied by each term.

[0038] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously. At the same time, other operations may be added to these processes, or one or more operations may be removed from these processes.

[0039] The forklift scale usage mode recognition method of the present application is applicable to any type of electronic forklift scale, and the present application does not limit the model and function of the electronic forklift scale. The electronic forklift scale can be manually pushed, or can be set on a motor vehicle, or have an automatic movement function.

[0040] Figure 1 1 is a schematic diagram of a side view of a forklift scale to which the use pattern recognition method according to an embodiment of the present application is applicable. Figure 1 As shown, the forklift scale 100 includes a fork tine 110, a junction box 120, a hydraulic cylinder 130, a terminal display instrument 140 and a lifting handle 150. The forklift scale 100 generally has two fork tine 110, each of which is provided with at least one pressure sensor 111. A lithium battery 121 and an acceleration sensor 122 are provided in the junction box 120.

[0041] In some embodiments, the acceleration sensor 122 includes a three-axis acceleration sensor.

[0042] The present application does not limit whether the forklift scale 100 itself is equipped with an acceleration sensor.

[0043] In one embodiment, the forklift scale 110 has only one acceleration sensor 122 for the pattern recognition method of the present application.

[0044] Figure 1 The illustrations are merely examples and are not intended to limit the specific structure of the forklift scale, the installation position of the acceleration sensor, etc.

[0045] Figure 2 is an exemplary flow chart of a method for using a pattern recognition method according to an embodiment of the present application. Figure 2 As shown, the usage pattern recognition method of this embodiment includes the following steps:

[0046] Step S210: defining a usage mode of the forklift scale, the usage mode including at least a static mode and a motion mode, the motion mode including a dragging mode and / or a lifting mode;

[0047] Step S220: collecting first data in each use mode, where the first data comes from an acceleration sensor provided on the forklift scale, wherein the forklift scale is in at least one use mode at the same time;

[0048] Step S230: preprocessing the first data to obtain second data;

[0049] Step S240: performing feature value extraction processing on the second data to obtain feature values;

[0050] Step S250: training a classifier according to the feature values ​​and the corresponding usage patterns; and

[0051] Step S260: using the trained classifier to identify the usage mode of the forklift scale according to the current feature value of the forklift scale.

[0052] The above steps S210 to S260 are described in detail below with reference to the accompanying drawings.

[0053] In step S210, the usage mode of the forklift scale 100 can be defined according to actual conditions. The usage mode includes at least a static mode and a motion mode, and the motion mode can be a dragging mode, or a lifting mode, or both. Among them, the static mode means that the forklift scale 100 does not move, and the fork tines 110 do not move. The dragging mode means that the forklift scale 100 is in the process of moving as a whole driven by the wheels. The lifting mode means that the fork tines 110 are lifted upward, and the forklift scale 100 can be in a fixed position or in the process of dragging. The dragging mode means that the forklift scale 100 is in motion. The lifting mode means that the fork tines 110 are in the process of lifting.

[0054] According to step S220, when the forklift scale 100 is in each working mode, the first data of the forklift scale 100 is collected, that is, the data acquired in real time by the acceleration sensor 122. This application does not limit the data volume and data format of the first data. There is no restriction on whether the data volume of the first data in different working modes is equal.

[0055] In some embodiments, in different working modes, the same amount of first data is collected as different data groups.

[0056] In some cases, if the forklift scale 100 is moving while the forks 110 are lifted upward, the forklift scale 100 may be considered to be in the drag mode and the lift mode at the same time.

[0057] In some embodiments, the situation where the forklift scale 100 is in the drag mode and the lift mode at the same time is defined as a separate working mode, ie, the drag lift mode.

[0058] In some embodiments, the working mode of the forklift scale 100 also includes a descending mode, that is, a mode in which the fork tines 110 move downward. Usually, when the fork tines 110 descend, the forklift scale 100 is in a fixed position, that is, not in a dragging mode.

[0059] In some embodiments, a weighing sensor is also provided on the forklift scale 100 of the present application. The present application does not limit the type, position, and quantity of the weighing sensor. Figure 1 The weighing sensor may be a pressure sensor 111. The weight of the items loaded on the forklift scale 100 may be obtained according to the weighing sensor, that is, weighing data may be obtained. According to these embodiments, the working modes of the forklift scale 100 may also be defined to include an empty mode and a loaded mode.

[0060] Figure 3 Schematic diagram of various usage modes that can be identified by the usage mode identification method according to an embodiment of the present application. Figure 3 As shown, the load mode and the no-load mode can be identified by the sensing results of the weighing sensor, and the working modes that can be identified according to the sensing results of the acceleration sensor include five modes: static mode, lifting mode, dragging mode, dragging and lifting mode, and descending mode. Combining these two working modes, the use mode recognition method of the present application can identify at least 10 modes, namely, static, lifting, dragging, dragging and lifting, and descending in the load mode, and static, lifting, dragging, dragging and lifting, and descending in the no-load mode.

[0061] Taking into account practical applications and safety identification, in some embodiments, the usage modes of the forklift scale include static mode, no-load lifting mode, no-load lowering mode, no-load dragging mode, no-load dragging lifting mode, loaded dragging mode, load lifting mode and load lowering mode, a total of eight usage modes.

[0062] In some embodiments, when the weighing data is within a preset weight range, the use mode is considered to be a no-load mode, otherwise, the use mode is a load mode. Ideally, the preset weight range is equal to 0, and accordingly, when the weighing data is greater than 0, it is considered to be a load mode. According to actual conditions, the preset weight range can be less than a preset error value, such as less than 10g, etc. This can avoid errors caused by some minor disturbances.

[0063] In step S230, the purpose of preprocessing the first data is to remove noise in the first data, for example, to remove some obviously unusable noise data, so that the preprocessed data is more suitable for subsequent pattern recognition algorithms. Preprocessing can be performed by using commonly used preprocessing methods in the art, including filtering, smoothing, etc.

[0064] In some embodiments, the step of preprocessing the first data includes:

[0065] Step S231: remove the first data exceeding the first threshold range as outliers, wherein the first threshold range is μ±3δ in a normal distribution, where μ is the mean and δ is the standard deviation.

[0066] According to step S231, some outliers in the original first data, that is, some singular points with large deviations from the mean, can be removed. It can be considered that these singular points are not obtained under the normal working mode, and the deviation of the data may be caused by special circumstances.

[0067] In some embodiments, the step of preprocessing the first data further includes, after step S231:

[0068] Step S232: Use the following formula to supplement the missing data after removing outliers:

[0069]

[0070] Among them, X miss is the missing data that needs to be supplemented, K is the total amount of data in the current group, and X miss-i and X miss+i X miss The previous set of data and the next set of data.

[0071] Steps S231 and S232 may be processed based on a set of data, where the set of data is, for example, a set of first data collected by an acceleration sensor within a certain period of time in a certain working mode.

[0072] After step S231 and step S232, the amount of data can be kept constant while improving the quality of the data. In particular, in one embodiment of the present application, the classifier used in step S250 is a support vector machine classifier based on wavelet transform. Therefore, in the preprocessing step of step S230, it also includes:

[0073] Step S233: Determine the length M of the time window corresponding to each usage mode and the number N of the consecutive moving time windows. M and N are used for subsequent wavelet decomposition of the acceleration data.

[0074] In step S240, the present application does not limit the specific feature value extraction processing method, and various feature extraction algorithms in the field of machine learning can be used to process the second data.

[0075] In particular, in order to reduce the cost of sensors and improve the utilization rate of sensors, the method for recognizing the usage pattern of forklift scales of the present application sets a separate three-axis acceleration sensor on the forklift scale 100. Therefore, the first data and the second data are both three-axis accelerations from the three-axis acceleration sensor. Accordingly, in step S240, the following steps are used to perform feature value extraction processing:

[0076] Step S241: performing multi-resolution wavelet transform on the three-axis acceleration to obtain wavelet coefficients, where the wavelet coefficients include high-frequency detail coefficients and low-frequency approximation coefficients;

[0077] Step S242: taking the square sum of the wavelet coefficients as the wavelet energy;

[0078] Step S243: selecting the peak value of the low-frequency approximation coefficient corresponding to the predetermined frequency range as the wavelet peak;

[0079] Step S244: The characteristic values ​​include wavelet energy and wavelet peak.

[0080] Specifically, in step S241, assuming that the three-axis acceleration is A(n)=[a x (n),a y (n),a z (n)], the formula of wavelet decomposition is as follows:

[0081]

[0082] Among them, d jk is the high-frequency detail coefficient, which indicates the fast-changing details in the signal. jk is a low-frequency approximate function, which indicates the overall trend of slow changes in the signal. j is a variable representing the scale, k is a variable representing time, and J is the number of decomposition levels. is a wavelet function, defined as

[0083] Wavelet function is a form of wavelet basis. Wavelet basis is a set of orthogonal functions with compact support, which can be obtained by translating and scaling a basic wavelet function. The basic wavelet function is usually called "mother wavelet" or "prototype wavelet". is a scaling function, defined as The scaling function is used to help determine the scale and translation parameters of the wavelet basis, thereby affecting the resolution and sensitivity of the wavelet transform.

[0084] In one embodiment of the present application, the db5 wavelet is selected as the mother wavelet to decompose the acceleration data.

[0085] In some embodiments, step S242 is specifically: taking the sum of the squares of the high-frequency detail coefficients of the third to seventh layers and the sum of the squares of the low-frequency approximation coefficients of the seventh layer as the wavelet energy. Further, step S242 is: calculating the wavelet energy using the following formula:

[0086]

[0087] Wherein, j is the variable of the telescopic scale, and k is the number of wavelet coefficients after decomposition. Formula (3) consists of two parts, the first part is the sum of the squares of the high-frequency detail coefficients from the third layer to the seventh layer, and the second part is the square of the low-frequency approximation coefficient of the seventh layer. The sum of the first part and the second part is used as the wavelet energy, which is proposed by the present application for the recognition of the working mode of the forklift scale.

[0088] In some embodiments, the corresponding wavelet layer number is determined according to a predetermined frequency range. The inventors of the present application have found through experiments that the actual movement frequency range of the forklift scale is about 0.01 Hz. The sampling frequency of data acquisition in step S220 of the present application is 20 Hz. Therefore, after the wavelet change, the frequency corresponding to the seventh layer of wavelet signal is about 0.01 Hz, which is similar to the range of the actual movement frequency of the forklift scale. Relatively speaking, the frequency of the useful signal used to reflect the usage mode of the forklift scale is low, and the high-frequency components in the signal are considered to be noise. Therefore, step S243 is specifically: select the peak value of the low-frequency approximation coefficient of the seventh layer as the wavelet peak. This is proposed by the inventors of the present application to solve the technical problem. The inventors of the present application have found through experiments that by using the wavelet energy calculated by step S242 and the wavelet peak calculated by step S243 as the characteristic value, the accuracy of the forklift scale working mode recognition can be significantly improved, and the specific results will be described later.

[0089] It should be noted that in other embodiments, the number of layers used in step S242 and step S243 may be determined based on the signal sampling frequency in step S220 and may not be limited to the above-mentioned number of layers.

[0090] Further, in some embodiments, after the wavelet energy is obtained in step S242, the result generated by combining the wavelet energy with the interquartile range is used as the eigenvalue corresponding to the wavelet energy. The interquartile range is defined as "The difference between the 75th and 25th percentiles of the data", that is, the value at 75% of a set of data minus the value at 25%.

[0091] In some embodiments, after the wavelet peak is obtained in step S243, a result generated by combining the wavelet peak with the interquartile range is used as a characteristic value corresponding to the wavelet peak.

[0092] In some embodiments, the wavelet energy and the wavelet peak are processed simultaneously, and the results generated by combining them with the interquartile range are used as feature values ​​for subsequent classifier input.

[0093] In some embodiments, after the features are obtained in step S240, a Wrapper feature subset selection (FSS) method is further used to perform dimensionality reduction processing on the feature values.

[0094] The Wrapper feature subset selection method includes the following steps:

[0095] Step S244: Feature selection search.

[0096] Performing feature selection search requires a state space (Status Space), an initial state (InitialStatus), a termination condition (Termination Condition) and a search mechanism (Search Engine). This application regards each working mode as a state, and each state represents a feature subset. For n features, each state has n bits (bit), and each bit indicates whether the feature is retained (present) or discarded (absent). If retained, the bit is recorded as 1, otherwise it is recorded as 0. Adding or deleting a feature from a state is an operation (operator). For n features, the size of the search space is 0(2n). When n is relatively large, this search space is relatively large, and it is impractical to search the entire space. Therefore, this application selects the best-first search mechanism, starts forward search (search forward) from an empty feature set, and the termination condition is 5 backtrackings (backtracking).

[0097] Step S245: Evaluate whether the feature subset is optimal in combination with the learning algorithm.

[0098] In some embodiments, the use pattern recognition method of the present application adopts recognition rate estimation (accuracyestimation) as a heuristic function (heuristic function) and an evaluation function (evaluation function). The recognition rate estimation method selects a five-fold cross-validation method (five-fold cross-validation), which is repeated multiple times for each feature. The number of repetitions is determined by the standard deviation of the recognition rate estimation. The learning algorithm used is a support vector machine (SVM). The present application uses a one-to-one strategy of SVM and uses a "Max-Wins" voting strategy to generate an output classification result.

[0099] refer to Figure 2 In step S250, in the process of training the classifier based on the feature values ​​obtained in the previous steps and the usage pattern corresponding to the group of data, the above S244-S245 can be used to evaluate the recognition effect of the classifier.

[0100] In some embodiments, step S250 further includes estimating the recognition rate of the classifier using a recognition rate estimation algorithm, and updating the model parameters of the classifier according to the estimation result.

[0101] refer to Figure 1 In step S260, after the classifier has been trained, the current feature value can be extracted according to the current first data of the forklift scale 100, and the usage mode of the forklift scale can be identified according to the current feature value, so that accurate results can be obtained conveniently and quickly.

[0102] Figure 4 FIG. 1 is an exemplary flow chart of a method for recognizing a usage pattern of a forklift scale according to an embodiment of the present application. Figure 1 The embodiment shown in the figure is refined. Figure 4 As shown, the usage pattern recognition method of this embodiment includes the following steps:

[0103] Step S410: Data collection, which is equivalent to the step S210 described above.

[0104] According to this embodiment, the usage patterns to be classified are defined in advance.

[0105] Step S412: Data preprocessing, which is equivalent to the above-mentioned step S230, and may include steps S231 to S233.

[0106] Step S414: extracting wavelet energy and wavelet peaks by wavelet decomposition, which is equivalent to the aforementioned step S240, extracting characteristic values, and may include steps S241 to S244.

[0107] Step S416: The wavelet energy is combined with the interquartile range to generate a characteristic value, and the wavelet peak is combined with the interquartile range to generate a characteristic value.

[0108] Step S418: Wrapper feature subset selects dimension reduction feature values, which is equivalent to steps S244 to S245 described above.

[0109] Step S420: Obtain a trained SVM classifier.

[0110] Step S422: Determine the confidence Conf of the SVM classifier. If Conf>a, execute step S242; otherwise, return to execute S418. Where a is a preset empirical value. The confidence can be automatically obtained by the design tool of the classifier.

[0111] Step S424: Output the recognition result.

[0112] When Conf≤a, it means that the classification result of the current SVM classifier is not ideal, and the above step S418 is repeated to optimize the eigenvalues, and the SVM classifier is retrained using the optimized eigenvalues.

[0113] By adopting steps S410 to S424, the usage mode of the forklift scale 100 can be classified according to simple acceleration data to obtain an accurate classification result.

[0114] The inventors of this application compared the SVM classification algorithm based on wavelet transform of this application with the classification algorithm based on time domain and the classification algorithm based on frequency domain. The results are as follows: Figure 5A to Figure 5C As shown. The following Table 1 shows the names of the time domain features used by the time domain-based classification algorithm and their corresponding calculation formulas. The following Table 2 shows the names of the frequency domain features used by the frequency domain-based classification algorithm and their corresponding calculation formulas.

[0115] Table 1: Time domain characteristics and their corresponding calculation formulas

[0116]

[0117]

[0118] Table 2: Frequency domain characteristics and their corresponding calculation formulas

[0119]

[0120] Figure 5AIt is the result obtained by classification using the time domain features shown in Table 1. The vertical axis represents the actual usage pattern, and the horizontal axis represents the predicted usage pattern output by the classifier. The horizontal axis from left to right and the vertical axis from top to bottom are lifting up, moving and lifting, moving by human, moving at variable speeds, putting down, and static. The numbers in the squares on the diagonal from the upper left to the lower right represent the consistency between the actual category and the predicted category, between 0 and 1, with the number 1 indicating complete consistency and the number 0 indicating complete inconsistency. Figure 5A As shown in the figure, the result is an indicator for evaluating classification performance: normalized confusion matrix.

[0121] Figure 5B It is the result obtained by classification using the frequency domain features shown in Table 2. Figure 5C It is the result obtained by using the pattern recognition method of the present application for classification. Figure 5B and Figure 5C The meaning of the horizontal and vertical axes is the same as Figure 5A Same. Figure 5C As shown, 3 of the diagonal squares are 1, and the other two categories are 0.89 and 0.97, which are relatively high. Overall, Figure 5C The classification result shown is obviously the best among the three methods, indicating that the method of the present application has the highest accuracy in identifying the usage pattern of the forklift scale.

[0122] The present application also includes a forklift scale usage mode recognition device, including a memory and a processor, wherein the memory is used to store instructions executable by the processor; the processor is used to execute the instructions to implement the forklift scale usage mode recognition method described above.

[0123] Figure 6 1 is a system block diagram of a forklift scale usage pattern recognition device according to an embodiment of the present application. Figure 6As shown, the usage pattern recognition device 600 may include an internal communication bus 601, a processor 602, a read-only memory (ROM) 603, a random access memory (RAM) 604, and a communication port 605. When applied on a personal computer, the usage pattern recognition device 600 may also include a hard disk 606. The internal communication bus 601 can realize data communication between the components of the usage pattern recognition device 600. The processor 602 can make judgments and issue prompts. In some embodiments, the processor 602 can be composed of one or more processors. The communication port 605 can realize data communication between the usage pattern recognition device 600 and the outside. In some embodiments, the usage pattern recognition device 600 can send and receive information and data from the network through the communication port 605. The usage pattern recognition device 600 may also include different forms of program storage units and data storage units, such as a hard disk 606, a read-only memory (ROM) 603 and a random access memory (RAM) 604, which can store various data files used for computer processing and / or communication, and possible program instructions executed by the processor 602. The processor executes these instructions to implement the main part of the method. The result processed by the processor is transmitted to the user device through the communication port and displayed on the user interface.

[0124] The above-mentioned forklift scale usage pattern recognition device can be implemented as a computer program, stored in the hard disk 606, and can be loaded into the processor 602 for execution to implement the forklift scale usage pattern recognition method of the present application.

[0125] The present application also includes a computer-readable medium storing a computer program code, which, when executed by a processor, implements the forklift scale usage pattern recognition method described above.

[0126] When the forklift scale usage pattern recognition method is implemented as a computer program, it can also be stored in a computer-readable storage medium as a product. For example, a computer-readable storage medium may include, but is not limited to, a magnetic storage device (e.g., a hard disk, a floppy disk, a magnetic strip), an optical disk (e.g., a compact disk (CD), a digital versatile disk (DVD)), a smart card, and a flash memory device (e.g., an electrically erasable programmable read-only memory (EPROM), a card, a stick, a key drive). In addition, the various storage media described herein can represent one or more devices and / or other machine-readable media for storing information. The term "machine-readable medium" may include, but is not limited to, wireless channels and various other media (and / or storage media) that can store, contain and / or carry code and / or instructions and / or data.

[0127] It should be understood that the embodiments described above are only illustrative. The embodiments described herein may be implemented in hardware, software, firmware, middleware, microcode, or any combination thereof. For hardware implementation, the processor may be implemented in one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), processors, controllers, microcontrollers, microprocessors, and / or other electronic units designed to perform the functions described herein, or a combination thereof.

[0128] Some aspects of the present application may be performed entirely by hardware, entirely by software (including firmware, resident software, microcode, etc.), or by a combination of hardware and software. The above hardware or software may be referred to as "data blocks", "modules", "engines", "units", "components" or "systems". The processor may be one or more application specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DAPDs), programmable logic devices (PLDs), field programmable gate arrays (FPGAs), processors, controllers, microcontrollers, microprocessors, or combinations thereof. In addition, various aspects of the present application may be expressed as computer products located in one or more computer-readable media, which include computer-readable program codes. For example, computer-readable media may include, but are not limited to, magnetic storage devices (e.g., hard disks, floppy disks, tapes ...), optical disks (e.g., compact disks CDs, digital versatile disks DVDs ...), smart cards, and flash memory devices (e.g., cards, sticks, key drives ...).

[0129] A computer-readable medium may include a propagated data signal containing computer program code, such as in baseband or as part of a carrier wave. The propagated signal may have a variety of manifestations, including electromagnetic, optical, etc., or a suitable combination. A computer-readable medium may be any computer-readable medium other than a computer-readable storage medium, which may be connected to an instruction execution system, device or apparatus to communicate, propagate or transmit a program for use. The program code on the computer-readable medium may be propagated via any suitable medium, including radio, cable, fiber optic cable, radio frequency signal, or similar medium, or any combination of the above mediums.

[0130] The basic concepts have been described above. Obviously, for those skilled in the art, the above invention disclosure is only used as an example and does not constitute a limitation of the present application. Although not explicitly stated herein, those skilled in the art may make various modifications, improvements and amendments to the present application. Such modifications, improvements and amendments are suggested in the present application, so such modifications, improvements and amendments still belong to the spirit and scope of the exemplary embodiments of the present application.

[0131] At the same time, the present application uses specific words to describe the embodiments of the present application. For example, "one embodiment", "an embodiment", and / or "some embodiments" refer to a certain feature, structure or characteristic related to at least one embodiment of the present application. Therefore, it should be emphasized and noted that "one embodiment" or "an embodiment" or "an alternative embodiment" mentioned twice or more in different positions in this specification does not necessarily refer to the same embodiment. In addition, some features, structures or characteristics in one or more embodiments of the present application can be appropriately combined.

[0132] In some embodiments, numbers describing the number of components and attributes are used. It should be understood that such numbers used in the description of the embodiments are modified by the modifiers "about", "approximately" or "substantially" in some examples. Unless otherwise specified, "about", "approximately" or "substantially" indicate that the numbers are allowed to vary by ±20%. Accordingly, in some embodiments, the numerical parameters used in the specification and claims are approximate values, which may change according to the required features of individual embodiments. In some embodiments, the numerical parameters should take into account the specified significant digits and adopt the general method of retaining digits. Although the numerical domains and parameters used to confirm the breadth of their range in some embodiments of the present application are approximate values, in specific embodiments, the setting of such numerical values ​​is as accurate as possible within the feasible range.

Claims

1. A method for identifying a usage pattern of a forklift scale, characterized in that: include: Defining a usage mode of the forklift scale, wherein the usage mode includes at least a stationary mode and a motion mode, and the motion mode includes a dragging mode and / or a lifting mode; Collecting first data in each of the usage modes, the first data coming from an acceleration sensor disposed on the forklift scale, wherein the forklift scale is in at least one usage mode at the same time; Preprocessing the first data to obtain second data; Performing feature value extraction processing on the second data to obtain feature values; Training a classifier based on the feature values ​​and corresponding usage patterns; and The trained classifier is used to identify the usage mode of the forklift scale according to the current feature value of the forklift scale.

2. The usage pattern recognition method according to claim 1, characterized in that: The forklift scale is also provided with a weighing sensor, which is used to obtain the weighing data of the forklift scale. The usage mode also includes a no-load mode and a load mode. When the weighing data is within a preset weight range, the usage mode is the no-load mode, otherwise, the usage mode is the load mode.

3. The usage pattern recognition method according to claim 1, characterized in that: The movement modes also include a descending mode and / or a drag-and-lift mode.

4. The usage pattern recognition method according to claim 1, characterized in that: The step of preprocessing the first data includes: The first data exceeding a first threshold range are removed as outliers.

5. The usage pattern recognition method according to claim 4, characterized in that: The first threshold range is μ±3δ in a normal distribution, where μ is the mean and δ is the standard deviation.

6. The usage pattern recognition method according to claim 4, characterized in that: The step of preprocessing the first data further includes: The following formula is used to supplement the missing data after removing outliers: Among them, X miss is the missing data that needs to be supplemented, K is the total amount of data in the current group, X miss-i and X miss+i X miss The previous set of data and the next set of data.

7. The usage pattern recognition method according to claim 4, characterized in that: The classifier includes a support vector machine classifier based on wavelet transform, and the step of preprocessing the first data also includes: The length of the time window corresponding to each of the usage modes and the number of the consecutive moving time windows are determined.

8. The usage pattern recognition method according to claim 7, characterized in that: The acceleration sensor includes a three-axis acceleration sensor, the second data includes three-axis acceleration, and the step of performing feature value extraction processing on the second data includes: Performing a multi-resolution wavelet transform on the three-axis acceleration to obtain wavelet coefficients, wherein the wavelet coefficients include high-frequency detail coefficients and low-frequency approximation coefficients; Taking the square sum of the wavelet coefficients as wavelet energy; Selecting the peak value of the low-frequency approximation coefficient corresponding to the predetermined frequency range as the wavelet peak; The characteristic values ​​include the wavelet energy and the wavelet peak.

9. The usage pattern recognition method according to claim 8, characterized in that: The result generated by combining the wavelet energy with the interquartile range is used as the eigenvalue corresponding to the wavelet energy.

10. The usage pattern recognition method according to claim 8, characterized in that: The step of taking the sum of the squares of the wavelet coefficients as wavelet energy includes: taking the sum of the squares of the high-frequency detail coefficients of the third to seventh layers and the sum of the squares of the low-frequency approximation coefficients of the seventh layer as the wavelet energy; the step of selecting the peak value of the low-frequency approximation coefficient corresponding to a predetermined frequency range as the wavelet peak includes: selecting the peak value of the low-frequency approximation coefficient of the seventh layer as the wavelet peak.

11. The usage pattern recognition method according to claim 8, characterized in that: The result generated by combining the wavelet peak with the interquartile range is used as the eigenvalue corresponding to the wavelet peak.

12. The usage pattern recognition method according to claim 1, characterized in that: The step of performing feature value extraction processing on the second data also includes: after obtaining the feature value, using a Wrapper feature subset selection method to perform dimensionality reduction processing on the feature value.

13. A usage mode recognition device for a forklift scale, comprising: a memory for storing instructions executable by a processor; A processor, configured to execute the instructions to implement the method according to any one of claims 1 to 12.

14. A computer readable medium storing computer program code, wherein the computer program code, when executed by a processor, implements the method according to any one of claims 1 to 12.

Images