Lossy data compression algorithm based on equipment operation condition

Abstract

The invention belongs to the technical field of equipment intelligent operation and maintenance, and particularly relates to a lossy data compression algorithm based on an equipment operation condition, original data acquired by a vibration sensor and converted by an analog-to-digital converter is a float-type array, each float-type data in the array occupies 4 bytes, 6-bit effective data after a decimal point is ensured, and the lossy data compression efficiency is improved. The method comprises the following steps of: multiplying an original array by 6 power of 10, rounding to obtain an array of a new integer type, averaging the array of the new integer type, and reserving an integer part to obtain a zoom coefficient; dividing the new array by the scaling coefficient to obtain a scaled array; and removing high-order two bytes in int elements in the zoomed array to form a compressed short type array, and outputting the short type array as a lossy compression result. Under the condition that the analysis quality is guaranteed, the transmitted data volume is reduced to half of the original data volume, and the effect that the communication feedback speed is higher is effectively improved.

Description

Technical Field

[0001] The present invention belongs to the technical field of equipment intelligent operation and maintenance, and specifically relates to a lossy data compression algorithm based on equipment operating conditions. Background Art

[0002] Existing wireless vibration sensor data compression methods are divided into lossy compression and lossless compression. The analysis of vibration sensor data involves accuracy issues. When meeting the analysis accuracy requirements, some of the original data accuracy can be appropriately lost. The compression rate of lossless compression data is low, and the compression rate of lossy compression can be adjusted according to the different accuracy losses, but the compression effect that can be achieved is still limited. The power consumption loss of the sensor mainly exists in two parts, one is the static loss during the operation of the sensor, and the other is the RF power consumption during the RF transmission process. The more data bytes there are, the greater the RF power consumption is under the same RF power, and the shorter the operating life of the sensor. Summary of the invention

[0003] The purpose of the present invention is to provide a lossy data compression algorithm based on the operating conditions of the equipment, in order to reasonably compress the data collected by the wireless vibration sensor to reduce the radio frequency transmission time, thereby achieving the purpose of reducing the power consumption of the sensor and extending the service life.

[0004] The lossy data compression algorithm based on the equipment operation condition comprises the following steps: the original data acquired by the vibration sensor and converted by the analog-to-digital converter is a float type array, and each float type data in the array occupies 4 bytes to ensure the valid data of 6 decimal places; a new integer type array is obtained by multiplying the original array by 10 to the sixth power and then rounding; the new integer type array is averaged and the integer part is retained to obtain a scaling factor; the scaled array is obtained by dividing the new array by the scaling factor; the high-order two bytes of the four-byte int element in the scaled array are removed to form a compressed short type array, and the short type array is output as the result of lossy compression.

[0005] Preferably, the data collected by the vibration sensor is converted by an analog-to-digital converter to obtain raw data, the fluctuation amplitude of the raw data is calculated, and the fluctuation amplitude is compared with the set fluctuation threshold. When the fluctuation amplitude exceeds the fluctuation threshold, it indicates that the equipment is degraded. A lossless compression method is used to compress the data and upload it to the gateway for analysis of the equipment. When the fluctuation amplitude does not exceed the fluctuation threshold, a low-precision lossy compression algorithm is used to transmit data to the gateway.

[0006] Preferably, for each array, the average of the part of the data greater than 0 is taken to obtain the average fluctuation of the positive direction operation of the device; when the average fluctuation of the positive direction is stable, that is, the average fluctuation of the positive direction operation does not exceed the fluctuation threshold.

[0007] Preferably, the size of the scaled array is between -32768 and 32767 and is an array of int type.

[0008] Preferably, the amount of data transmitted is reduced to half of the original amount, the time complexity of the compression algorithm is O(1), and parallel computing is supported by a GPU.

[0009] Preferably, the vibration sensor is used in steel metallurgy scenarios.

[0010] Preferably, the data compressed by the lossy data compression algorithm is further compressed by other compression algorithms to reduce the data volume to less than 30% of the original data volume.

[0011] The present invention has the following advantages: This method reasonably compresses the data collected by the wireless vibration sensor by improving the lossy data compression algorithm, thereby reducing the radio frequency transmission time, reducing the sensor power consumption and extending the service life. This method is aimed at the business scenario of iron and steel metallurgy, and does not consider high-order data. Although it loses some accuracy, it can still retain the original data characteristics well, thereby reducing the amount of transmitted data to half of the original amount while ensuring the quality of analysis, effectively improving the effect of faster communication feedback speed. BRIEF DESCRIPTION OF THE DRAWINGS

[0012] Figure 1 The present invention is a flow chart of a lossy data compression algorithm based on equipment operating conditions. DETAILED DESCRIPTION

[0013] The specific implementation modes of the present invention are further explained in detail below by describing the embodiments with reference to the accompanying drawings, so as to help those skilled in the art to have a more complete, accurate and in-depth understanding of the inventive concept and technical solution of the present invention.

[0014] The vibration data of equipment in the steel industry has its own characteristics, that is, when the equipment is running stably and well, its vibration range is mostly stable within a range; as long as the equipment has not deteriorated and there is no external influence, the collected data are similar, and the analysis characteristics of each set of data (spectrum, envelope, etc.) are close to the same.

[0015] like Figure 1As shown, the present invention provides a lossy data compression algorithm based on the equipment operation condition, including the following steps: the vibration sensor collects data and converts it through an analog-to-digital converter (ADC) to obtain raw data, calculates the fluctuation amplitude of the raw data, compares the fluctuation amplitude with the set fluctuation threshold, and when the fluctuation amplitude exceeds the fluctuation threshold, it indicates that the equipment is deteriorated, and uses a lossless compression method to compress the data and upload it to the gateway for analysis of the equipment, and when the fluctuation amplitude does not exceed the fluctuation threshold, uses a low-precision lossy compression algorithm to transmit data to the gateway. This can reduce the RF transmission time, thereby achieving the purpose of reducing sensor power consumption and extending service life.

[0016] During the stable operation of the device, the vibration sensor acquires data through ADC as a float type array, and each float type data in the array occupies 4 bytes. For each array, take the average of the data greater than 0 to obtain the average fluctuation of the device in the positive direction; when the average fluctuation in the positive direction is stable, that is, when the average fluctuation in the positive direction does not exceed the fluctuation threshold, use a low-precision lossy compression algorithm to compress the data and transmit the data to the gateway; otherwise, use a lossless compression method to compress the data and transmit the data to the gateway.

[0017] This algorithm improves the low-precision lossy compression algorithm to obtain a new lossy data compression algorithm, which specifically includes: float type data can guarantee valid data of 6 decimal places, and a new integer (int type) array is obtained by multiplying the original array by 10 to the sixth power and rounding it. Then, the new integer array is averaged to retain the integer part to obtain the scaling factor; the scaled array is obtained by dividing the new array by the scaling factor. The size of the scaled array should be between -32768 and 32767, and it is an array of int type. Each element in the scaled array occupies 4 bytes. The high two bytes of the four-byte int element in the scaled array are removed to form a compressed short type array, so as to achieve the effect of lossy compression of the data and retaining 4 decimal places of precision.

[0018] A normal data of type int is represented by 4 bytes. For example, in binary, a data of type int is 0x0000 0000 0000 0001. Based on the scenario of iron and steel metallurgy, in most cases, the effective bits collected by the vibration sensor are only the last few bits, and the front bits are all 0. Therefore, this method does not consider the high-order data, removes the high-order two bytes in front of the data of type int, and only uses the low-order two bytes of the last two bits to represent each number, such as 0x 0000 0001.

[0019] The above steps can be specifically expressed as:

[0020] scaled_data=np.floor(data*10**6).astype(int);

[0021] mean_value=int(np.mean(scaled_data));

[0022] new_array=np.floor(scaled_data / mean_value).astype(int);

[0023] short_array=new_array.astype(np.int16);

[0024] Among them, scaled_data represents the scaled data, np.floor() represents rounding down, data represents the initial data, astype(int) represents the conversion of the variable data type to int type, mean_value represents the average value of the array elements, int() represents the conversion to an integer; new_array represents the created array, short_array represents an array of short type, np.int16 represents the conversion to a 16-bit 4-byte integer type object, and astype(np.int16) represents the conversion of the variable data type to a 16-bit 4-byte int type.

[0025] Compared with the existing lossy compression method, the above new algorithm will reduce the amount of data that needs to be transmitted to half of the original amount. If there is a GPU that supports parallel computing, the calculation will be faster because the time complexity of this compression algorithm is O(1), so it can respond quickly and maintain efficient performance. The disadvantage is that this algorithm is based on the business scenario of iron and steel metallurgy, and is not as universal as other compression algorithms. After using this algorithm, other algorithms can be used to further compress the compressed data and reduce the data volume to less than 30% of the original data volume.

[0026] The algorithm is simple, thus achieving a faster communication feedback speed; combined with the business needs of iron and steel metallurgy, the amount of data is compressed while ensuring the quality of analysis. At the same time, a floating point type of data, that is, float type data, is normally represented by 4 bytes. Based on the scenario of iron and steel metallurgy, in most cases its valid bits are only the last few bits. The front ones are all 0; therefore, this method does not consider the high-order data, and only uses two bytes to represent each number 0x00000001. Although this loses some precision, it can still retain the original data characteristics well.

[0027] After using this algorithm for data compression and transmission between the vibration sensor and the gateway, the RF power consumption is effectively reduced. When the equipment is in good working condition, a low-precision lossy compression algorithm is used for data compression, which can effectively extend the service life of the sensor.

[0028] The present invention is described above by way of example in conjunction with the accompanying drawings. It is obvious that the specific implementation of the present invention is not limited to the above-mentioned method. As long as various non-substantial improvements are made using the inventive concept and technical solution of the present invention, or the inventive concept and technical solution are directly applied to other occasions without improvement, they are all within the protection scope of the present invention.

Claims

1. A lossy data compression algorithm based on equipment operating conditions, characterized in that: The original data collected by the vibration sensor and converted by the analog-to-digital converter is a float type array, and each float type data in the array occupies 4 bytes to ensure the valid data of 6 decimal places. A new integer type array is obtained by multiplying the original array by 10 to the sixth power and rounding it to the integer. The new integer type array is averaged to retain the integer part to obtain the scaling factor; the scaled array is obtained by dividing the new array by the scaling factor; the high two bytes of the four-byte int element in the scaled array are removed to form a compressed short type array, and the short type array is output as the result of lossy compression.

2. The lossy data compression algorithm based on equipment operating conditions according to claim 1, characterized in that: The data collected by the vibration sensor is converted by the analog-to-digital converter to obtain the original data. The fluctuation amplitude of the original data is calculated and compared with the set fluctuation threshold. When the fluctuation amplitude exceeds the fluctuation threshold, it means that the equipment is deteriorating. The data is compressed using a lossless compression method and uploaded to the gateway for analysis of the equipment. When the fluctuation amplitude does not exceed the fluctuation threshold, a low-precision lossy compression algorithm is used to transmit data to the gateway.

3. A lossy data compression algorithm based on equipment operating conditions according to claim 1 or 2, characterized in that: For each array, take the average of the data greater than 0 to obtain the average fluctuation of the positive direction operation of the device; when the average fluctuation of the positive direction is stable, it means that the average fluctuation of the positive direction operation does not exceed the fluctuation threshold.

4. The lossy data compression algorithm based on equipment operating conditions according to claim 1, characterized in that: The size of the scaled array is between -32768 and 32767 and is an array of int type.

5. The lossy data compression algorithm based on equipment operating conditions according to claim 1, characterized in that: The amount of data transmitted is reduced to half of the original amount. The time complexity of the compression algorithm is O(1), and parallel computing is supported by GPU.

6. The lossy data compression algorithm based on equipment operating conditions according to claim 1, characterized in that: The vibration sensor is used in the context of steel metallurgy.

7. The lossy data compression algorithm based on equipment operating conditions according to claim 1, characterized in that: The data compressed by the lossy data compression algorithm is further compressed by other compression algorithms to reduce the data volume to less than 30% of the original data volume.

Images